Zohe Mustafa

I Grow Your User Base w/ AI, Automation + Vibe Coding. CMO Head of Marketing / Growth Hacking / GTM. AI Generalist, AI Strategy | No-Code Workflows - Development, SaaS Founder. AI Agents, Adverts, Avatars, & Automations.

𝗜’m a restless builder who thrives on creating, testing, and scaling ideas. From Growth Hakka to JournoPitch to AICRM, I treat every venture as a living organism, feeding it, tweaking it, and seeing how it grows. Over nearly three decades I’ve worked with global enterprises like Virgin Mobile, Nokia, and Accenture as well as ambitious startups, always with the same goal: turning creativity and systems thinking into growth.𝗪𝗵𝗮𝘁 𝘀𝗲𝘁𝘀 𝗺𝗲 𝗮𝗽𝗮𝗿𝘁 is how I connect dots others overlook. One hour I’m mapping a SaaS GTM strategy; then Vibe Coding an app, then next, mentoring students on how to turn complex analytics into stories. I dive into the granular, dashboards, frameworks, AI prompts, while never losing sight of the ecosystems and futures they serve.𝗜 𝘀𝗽𝗲𝗰𝗶𝗮𝗹𝗶𝘀𝗲 𝗶𝗻 𝗮𝗽𝗽𝗹𝘆𝗶𝗻𝗴 𝗔𝗜 𝗮𝗻𝗱 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 to transform go-to-market strategies, SaaS scalability, and customer acquisition. As founder of Growth Hakka, I help businesses rethink traditional approaches and build systems that are smart, scalable, and repeatable. I’ve launched ventures like JournoPitch and AICRM my first Vibe Coding projects, designed to compress tomorrow’s innovation into tools companies can use today.𝗣𝗹𝗮𝘆 𝗶𝘀 𝘄𝗼𝘃𝗲𝗻 𝗶𝗻𝘁𝗼 𝗺𝘆 𝘄𝗼𝗿𝗸. I’ve gamified complex ideas through frameworks because storytelling sticks where spreadsheets don’t. Beneath the creativity is discipline: structure, clarity, and measurable outcomes that drive results at scale.𝗘𝗮𝗿𝗹𝘆 𝘀𝗶𝗴𝗵𝘁. I’ve had a track record of spotting shifts before they hit. In 2012 I predicted voice would become the dominant input before Siri or Alexa mattered. In 2017 I concepted Autherbot, anticipating AI copywriting tools years before they went mainstream. Today I’m asking bigger questions: how will AI-human coexistence reshape marketing, PR, and business?𝗪𝗵𝗮𝘁 𝗱𝗿𝗶𝘃𝗲𝘀 𝗺𝗲 is connecting people, ideas, and technologies. My work lives at the junction of innovation and execution, with one purpose: to equip businesses and individuals with the tools, insights, and strategies to thrive in an AI-powered future.𝗛𝗼𝘄 𝗜 𝘄𝗼𝗿𝗸: diagnose bottlenecks, design experiments, automate for scale, measure what matters. I build repeatable growth loops in content, distribution, data, and iteration. I favour simple systems, blending no-code with code and treating AI as a co-pilot.𝗜𝗺𝗽𝗮𝗰𝘁: faster pipeline velocity, better conversion, lower CAC. With enterprises I’ve delivered multimillion-pound outcomes; with startups I help sharpen positioning and ship GTM playbooks.

Head of Marketing,  Head of Digital Marketing, CMO, CGO, Marketing Automation

Growth Hakka is a growth hacking consultancy based in London, UK. Worked on campaigns for these industry verticals:

SAAS providers, Telecoms, Logistics, News & Media, Hospitality, Technology, FinTech, Home & Garden, Social Networks, Health and Wellbeing & E-commerce stores.

I am a no-code, growth and digital marketing, and automation specialist

I join these companies and provide marketign leadership. I can slot in as aChief Growth Officer or Chief Marketing Officer, Head of Marketing or Head of Digital Marketing manage or build their existing teams. I provide leadership, vision, plan, execution strategy, and mentoring.

My raison d'etre revolves around the principle of growth hacking. What is growth hacking? The term originated in Silicon Valley and in its essence it’s the idea that an entrepreneur or organisation can take a non-traditional approach to increasing the growth rate/adoption of their product by “hacking” something together specifically for growth purposes. Its about achieving an objective rather than following a prescribed process. Its about the intersection between Marketing, Product, Technology, Data and even HR, Finance. 

Services whilst always focused on Growth, either as a managed service or strategic input include any relevant combinations of the below:

Affiliate Marketing
A/B Testing - Product and Marketing
Analytics, Dashboards 
Conversion Rate Optimisation, 
Content Marketing
CRM
Developer Marketing
Digital: Branding, Consumer Profiling
Digital Marketing & Merchandising
Digital Strategy
Digital Media Planning & Buying across Search, Display, Mobile & Social Media
E-Commerce
Email Marketing
Gamification Strategy
Influencer Marketing
Journalist Marketing
Marketing Automation 
Mobile Marketing & Advertising
Multi-Channel Strategy
Multi Screen Campaign Strategy
Platform / Tool Selection
PR Digital Distribution
Programmatic Advertising & Marketing
SEM & SEO
Social Media Automation, Care, Enterprise, Management, Marketing & Monitoring

𝗧𝗵𝗲 𝗳𝗮𝘀𝘁𝗲𝘀𝘁 𝘄𝗮𝘆 𝘁𝗼 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘆𝗼𝘂𝗿 𝗺𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴 𝗱𝗮𝘁𝗮? 𝗧𝗮𝗹𝗸 𝘁𝗼 𝗶𝘁.

For years, marketers learned to navigate data.
Hundreds of analytics tools…
Hours spent staring at the numbers until you spot a pattern.

The new layer of analytics? Conversational AI.

𝘠𝘰𝘶 𝘢𝘴𝘬 𝘢 𝘲𝘶𝘦𝘴𝘵𝘪𝘰𝘯.
𝘈𝘯𝘥 𝘵𝘩𝘦 𝘪𝘯𝘴𝘪𝘨𝘩𝘵 𝘢𝘯𝘴𝘸𝘦𝘳𝘴 𝘺𝘰𝘶 𝘣𝘢𝘤𝘬.

Recently, I tested this by chatting with my own Google Analytics data. I connected my GA4 account to Claude with Coupler.io and started getting insights straight away. 

I’m honestly impressed with how much time it can save you.
I got my answers in minutes.

Just one prompt 
and Claude returned not only the data, but the reasoning behind it.
What pages underperformed.
What might be hurting conversion.
What to fix first.

Slides below if you want to see how it works.
And the tool to set up the integration if you want to try it yourself:
https://lnkd.in/euQyYFxB

Claude GA4 and Coupler

𝗘𝘃𝗲𝗿 𝘁𝗿𝗶𝗲𝗱 𝘀𝗲𝘁𝘁𝗶𝗻𝗴 𝘂𝗽 𝗽𝗮𝘆𝗺𝗲𝗻𝘁𝘀 𝗶𝗻𝘀𝗶𝗱𝗲 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻 𝗮𝗽𝗽
and realised Stripe’s “simple” docs
aren’t that simple once you go beyond checkout?

That was me a few months ago.

Not theorising.
Actually building a real credit-based payment model
for JournoPitch, my AI PR platform.
Inside Replit.
From scratch.

No prebuilt templates.
No copy-paste tutorials.
Just Vibe Coding my way through it.

What do most people skip over
when they “add Stripe”?:

They stop at “taking payment.”
But a proper credit model needs to track usage,
deduct credits on API calls,
update balances in real time,
and sync that with Stripe events.

That’s a lot of moving parts.
So I broke it down into a flow, [3 pages below]
one that any solo builder can follow:

1️⃣ Define your credit logic
- how many credits per feature,
- and what happens when they hit zero.

2️⃣ Create Stripe Products and Prices
for each bundle (e.g. 100, 500, 1000 credits).

3️⃣ Build a webhook listener
that fires every time a payment succeeds
and updates the user’s credits in your database.

4️⃣ Map your API calls
so each one deducts the right number of credits.

5️⃣ Finally, wire up a dashboard
showing remaining credits,
recent transactions,
and top-up options.

That’s it.
Not “set up payments.”
𝗦𝗲𝘁 𝘂𝗽 𝗮 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗺𝗼𝗱𝗲𝗹.

When you understand how value flows through your app,
you’re not just building software.

𝘠𝘰𝘶’𝘳𝘦 𝘥𝘦𝘴𝘪𝘨𝘯𝘪𝘯𝘨 𝘢 𝘴𝘺𝘴𝘵𝘦𝘮 𝘵𝘩𝘢𝘵 𝘴𝘶𝘴𝘵𝘢𝘪𝘯𝘴 𝘪𝘵𝘴𝘦𝘭𝘧.

The Payment Flow Nobody Teaches You

𝗕𝘂𝗶𝗹𝗱 𝘀𝗺𝗮𝗹𝗹 𝘀𝗵𝗶𝗽 𝗳𝗮𝘀𝘁.
𝘉𝘦𝘤𝘢𝘶𝘴𝘦 𝘮𝘰𝘴𝘵 𝘱𝘦𝘰𝘱𝘭𝘦 𝘴𝘱𝘦𝘯𝘥 𝘮𝘰𝘯𝘵𝘩𝘴 𝘱𝘰𝘭𝘪𝘴𝘩𝘪𝘯𝘨 𝘢 𝘣𝘰𝘢𝘵 𝘵𝘩𝘢𝘵 𝘯𝘦𝘷𝘦𝘳 𝘭𝘦𝘢𝘷𝘦𝘴 𝘵𝘩𝘦 𝘣𝘦𝘢𝘤𝘩. Last night I re-read the 𝗟𝗼𝗿𝗲𝗹𝗶𝗴𝗵𝘁 𝗳𝗼𝘂𝗻𝗱𝗲𝗿'𝘀 𝘀𝗵𝘂𝘁𝗱𝗼𝘄𝗻 𝗮𝗿𝘁𝗶𝗰𝗹𝗲.

It hit harder than I expected. Google it. Very interesting read.

Not because he failed.
But because he said the quiet part out loud.

He said there is no magic AI optimisation.
No hidden lever in ChatGPT.
No special trick for Claude or Perplexity.

The answer was always the same.
Create better content.
Build relationships with 𝗮𝘂𝘁𝗵𝗼𝗿𝗶𝘁𝗮𝘁𝗶𝘃𝗲 𝗽𝘂𝗯𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀.
Show genuine expertise.
Do good marketing.

Funny how the future always circles back to the fundamentals.

When I built JournoPitch, I didn’t focus on complicated GEO ideas.
I started with one painful truth I saw everywhere.

People weren’t struggling with AI search.
They were struggling with the exact same bottlenecks they faced in traditional PR.
No time.
No clarity.
No relationships.
No structure to turn their story into something a journalist might actually publish.

So instead of trying to “optimise for AI search”, I built for a real human problem.
Make it easier to create good PR content.
Make it easier to get it in front of the right publications.
Make it faster to turn expertise into something credible and visible.

That’s where the vibe coding part fits in.

The whole point of vibe coding is to build from pain points outward, not feature lists inward.

In my JournoPitch and AICRM chats with AI, the pattern is always the same.

Start with the customer’s discomfort.
What slows them down?
What feels chaotic?
What costs them opportunities?
What “should be obvious” but somehow never gets done?

Then translate that into a clean PRD.
Not a bloated one.
Just a sequence of pain points mapped to outcomes.

From that, build the tiniest version that delivers a win.
Then ship it.
Then layer micro improvements.
Then ship again.

That’s how I approached JournoPitch rollouts.
One micro win at a time.
A little pyramid of momentum.
Every layer built on real feedback rather than hypothetical needs.

Nothing in the Lorelight article contradicts that.
If anything, it reinforces it.

Tools come and go.
Algorithms evolve.
AI models shift.
But usefulness wins.
Real expertise wins.
Credible mentions win.
Relationships win.

If an app helps people do the work that actually builds a brand, it survives.
If it chases a trend that doesn’t solve a real problem, it doesn’t.

Build small ship fast.
But build it because someone is drowning, not because you like the colour of the sails.

Build Small Ship Fast

𝗧𝗵𝗲 𝗨𝗞 𝗺𝗶𝗴𝗵𝘁 𝗯𝗲 𝘁𝗵𝗲 𝗹𝗮𝘀𝘁 𝗽𝗹𝗮𝗰𝗲 𝘆𝗼𝘂 𝗲𝘅𝗽𝗲𝗰𝘁
𝘁𝗼 𝗯𝗲 𝘀𝗮𝗳𝗲 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗿𝗼𝗯𝗼𝘁 𝘁𝗮𝗸𝗲𝗼𝘃𝗲𝗿.

𝘉𝘶𝘵 𝘵𝘩𝘦 𝘥𝘢𝘵𝘢 𝘵𝘦𝘭𝘭𝘴 𝘢 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘵 𝘴𝘵𝘰𝘳𝘺.

New analysis across Europe shows the UK is
actually the #1 most resilient nation to
automation.

Only 30% of our workforce faces high risk.
One of the lowest figures in Europe.

Our population decline is small compared
to neighbours staring down massive drops
in working-age talent.

And we’ve kept a solid support ratio:
roughly 3 working-age adults per retiree.

Compare that to Spain, where 75% of
jobs are at risk.

Or Slovakia, losing nearly 9% of their
workforce by 2050.

Meanwhile Iceland shows what the future
really looks like:
99% of people score high on computer
skills.
That’s what resilience actually means.

The real paradox?
Automation might be the answer to the
demographic crisis everyone is ignoring.

Europe needs AI just to keep productivity
from collapsing.

So the real question is this:
Is your country preparing for the shift?
Or are we sleepwalking into a skills gap?

New data suggests that while 70% of jobs in some regions are at risk, demographics might be our saving grace. In a small break from my usual marketing

Open article: The AI Paradox: Why The UK Is Surprisingly Resilient to the Robot Takeover by Zohe Mustafa • 2 min read

The AI Paradox: Why The UK Is Surprisingly Resilient to the Robot Takeover

𝗧𝗵𝗲𝗿𝗲 𝗯𝗲 𝗹𝗼𝘁𝘀 𝗼𝗳 𝘁𝗮𝗹𝗸𝘀 𝗮𝗯𝗼𝘂𝘁 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝗳𝗲𝗮𝘁𝘂𝗿𝗲𝘀.
𝗩𝗲𝗿𝘆 𝗹𝗶𝘁𝘁𝗹𝗲 𝗰𝗵𝗮𝘁 𝗮𝗯𝗼𝘂𝘁 𝗳𝗹𝗼𝘄.

And flow is the real engine.

Before JournoPitch had AI PR agents,
before it could interview you with Voice-to-Content,
before it tracked media mentions in under 60 seconds,
before it auto analysed pitches for newsworthiness,
before it sent personalised outreach sequences,
before any of that existed…

I had a cig packet vision.

Literally a rough sketch on whatever was closest.
Homepage.
Signup.
Onboarding.
First action.
First moment of value.

Every app I have ever built has started like that.

Same for AICRM.
Before journeys,
before segmentation scoring,
before the UX overhaul in Replit…
it started with a messy drawing of how someone should enter, see value, and act.

Is Vibe Coding about speed?
Nah. It is not.
It is about clarity. [the speed does help of course]

A cig packet user flow forces you to answer the questions that actually matter:

Who is landing here?
What are they trying to do?
How fast can I give them something valuable?
Where will they drop off?
What is the very next page?
And the next?
And the next?

𝗬𝗼𝘂 𝗱𝗼 𝗻𝗼𝘁 𝗻𝗲𝗲𝗱 𝗮 𝗽𝗲𝗿𝗳𝗲𝗰𝘁 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲.
𝗬𝗼𝘂 𝗻𝗲𝗲𝗱 𝗮 𝘀𝘁𝗼𝗿𝘆 𝗼𝗳 𝗺𝗼𝘃𝗲𝗺𝗲𝗻𝘁.

JournoPitch’s real flow only became obvious when I sketched it badly first:

Home
→ Who are you
→ What are you trying to earn coverage for
→ How do we get your first pitch drafted
→ How do you build your first verified journalist list
→ How do you send it
→ How do we monitor the mentions
→ How do we help you respond fast
→ How do we automate your sequences
→ How do we help you build momentum

AICRM is the same story:

Home
→ Add first contact
→ Enrich
→ Score
→ Trigger first journey
→ Track the next step
→ Close the loop

All of it came from rough drawings.

This is why skipping user flows is the biggest silent killer of indie builds and MVPs.

You start building screens without knowing the story.
You start coding without knowing the journey.
You ship features with no path connecting them.

A builder needs a map.
Not an aesthetic.
A trajectory.

So if you are building something this week, try this.

Grab the worst piece of paper you can find.
Draw the user’s first 5 moves.
Forget UI.
Forget colours.
Forget components.

Focus on direction.

Then take that sketch and say to ChatGPT or Gemini:

Here is the flow.
Stress test it.
Tell me what is missing.
Tell me where users get stuck.
Tell me what you do not understand.
Turn this into a draft PRD.

Do that once and you will never go back.

𝘉𝘦𝘤𝘢𝘶𝘴𝘦 𝘵𝘩𝘦 𝘥𝘪𝘳𝘵𝘺 𝘴𝘬𝘦𝘵𝘤𝘩 𝘪𝘴 𝘵𝘩𝘦 𝘵𝘩𝘪𝘯𝘨 𝘵𝘩𝘢𝘵 𝘶𝘯𝘭𝘰𝘤𝘬𝘴 𝘵𝘩𝘦 𝘤𝘭𝘦𝘢𝘯 𝘣𝘶𝘪𝘭𝘥.

𝗠𝗼𝘀𝘁 𝗽𝗲𝗼𝗽𝗹𝗲 𝘁𝗿𝗲𝗮𝘁 𝗥𝗲𝗽𝗹𝗶𝘁 𝗹𝗶𝗸𝗲 𝗮 𝗰𝗼𝗱𝗶𝗻𝗴 𝘁𝗼𝗼𝗹.
𝗜 𝘁𝗿𝗲𝗮𝘁 𝗶𝘁 𝗹𝗶𝗸𝗲 𝗮 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗹𝗶𝗻𝗲.

When you’re building in public,
you learn quickly that speed is not
about typing faster.
Speed comes from reducing the number
of times you break your own momentum.

That’s what Replit gives you.

A single environment
where thinking, planning, testing,
refactoring, shipping, and fixing
all sit next to each other
instead of fighting for space.

I never realised how much energy I lost
jumping between tools
until I stopped doing it.

Replit became the place
where messy notes turn into user flows,
user flows turn into prompts,
prompts turn into working features,
and features turn into real products.

Not prototypes.
Not half-done experiments.
Actual, production grade apps
running live
with real users.

Both AICRM and JournoPitch
benefited from that shift.
Not because Replit wrote the apps for me,
but because it let me stay
in the same mental lane
from idea
to output
to iteration.

Everything feels tighter.
Cleaner.
More immediate.

Your loops compress.
Your feedback accelerates.
Your learning compounds.

You stop battling the workflow
and start building the product.

Replit eliminates
the need for a stack at all.
It becomes the stack.
P.S You can also build Agents with it. 

That simplicity creates confidence.
You take more risks.
You ship more frequently.
You break things without fear
because you can fix them
in the same place you built them.

If you’ve ever wondered
why people who vibe code
produce so much
so fast,
this is the reason.

It’s not magic.
It’s not shortcuts.
It’s environment design.
A workflow that stays out of the way
so your ideas can get out of your head
and into the world.

Replit for End to End

The Week Marketing Quietly Changed Shape
This week across my blog, seven different signals all pointed to one truth:
Marketing is being rewired faster than most teams realise.
AI search is taking power.
 Customer journeys are bypassing websites.
  Campaigns are becoming self-driving.
   AI is joining the team.
    And trust is becoming the new click. Or it always was?

I Grow Your User Base w/ AI, Automation + Innovation. CMO. Head of Marketing / Growth Hacking / GTM. AI Generalist, AI Strategy | No-Code Workflows - Development, SaaS Founder. AI Agents, Adverts, Avatars, & Automations.

One uncomfortable truth: The marketing ecosystem is rewiring itself faster than most teams realise. Workflows compressed. Search power shifted. Custom

Open article: The Week Marketing Quietly Changed Shape by Zohe Mustafa • 3 min read

𝗗𝗲𝗯𝘂𝗴𝗴𝗶𝗻 𝗶𝘀 𝗮 𝘀𝘁𝗮𝘁𝗲 𝗼𝗳 𝗺𝗶𝗻𝗱, 𝗻𝗼𝘁 𝗮 𝘀𝗸𝗶𝗹𝗹𝘀𝗲𝘁

I did not believe this early on.

When I first started Vibe Coding my apps, every bug felt like a personal attack.
A line of red. A vague error. Something I changed five minutes ago sabotaging something I built two days before. 

AAAARRRGGGHHHH!

It took a while to realise that debugging is not a technical exercise.
It is emotional regulation disguised as problem solving.

And once that clicked, everything got easier.

Not because the bugs suddenly got simpler.
But because I stopped treating them as emergencies.

I started treating them as conversations.

You build something with energy.
𝗧𝗵𝗲𝗻 𝗿𝗲𝗮𝗹𝗶𝘁𝘆 𝗿𝗲𝗽𝗹𝗶𝗲𝘀.
Sometimes gently.
Sometimes aggressively.
But always honestly.

And that honesty is where the growth happens.

When I’m working on JournoPitch or AICRM, the pattern is always the same.
A flow breaks. The AI gives me something half helpful.
Replit highlights twenty lines that all look equally wrong.

Old me would dive straight in, clicking around wildly, editing things instinctively, hoping to stumble across the answer.

𝘕𝘦𝘸 𝘮𝘦 𝘥𝘰𝘦𝘴 𝘵𝘩𝘦 𝘰𝘱𝘱𝘰𝘴𝘪𝘵𝘦.

I slow down.
Reproduce the issue.
Watch it carefully.
Ask better questions.
Guide the AI instead of surrendering to it.
And if I need to walk away, I do. [sometimes full of frustration]

Debugging calmly is not passive.
It’s controlled.

A deliberate choice to stay curious rather than panicked.

Because when you panic, you’re debugging the emotion.
When you’re calm, you debug the code.

And then the unlock becomes simple.

Every bug is a breadcrumb.
Every breadcrumb is a clue.
And every clue is trying to move you forward.

Not ruin your day. 
[Do check Replit threads on Reddit, can sometimes be amusing how angry people get]

So if you’re building in public, building with AI, building solo, or building with limited sleep and too much ambition, remember this:

You cannot rush clarity.
But you can create the conditions for it.

Take your time.
Observe.
Reframe.
Try again.

You will break things.
You will fix things.
And eventually, you will understand things you once thought were impossible.

𝗧𝗵𝗲 𝗺𝗼𝗺𝗲𝗻𝘁 𝗜 𝘀𝘁𝗼𝗽𝗽𝗲𝗱 “𝘄𝗶𝗻𝗴𝗶𝗻𝗴 𝗶𝘁”
𝗮𝗻𝗱 𝘀𝘁𝗮𝗿𝘁𝗲𝗱 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗣𝗥𝗗𝘀 [𝘗𝘳𝘰𝘥𝘶𝘤𝘵 𝘙𝘦𝘲𝘶𝘪𝘳𝘦𝘮𝘦𝘯𝘵𝘴 𝘋𝘰𝘤𝘶𝘮𝘦𝘯𝘵]
𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝗮𝗰𝗰𝗲𝗹𝗲𝗿𝗮𝘁𝗲𝗱.

When you’re vibe coding, it is so tempting
to jump straight into building.

Hit the editor.
Fire up the agent.
Spin up the repo.
And vibe your way into a working feature.

I learned a few frustrating truths the hard way.

Vibes are amazing for creativity.
Vibes are terrible for clarity.

The difference between two days of chaos
and two days of rapid progress
was whether I stopped long enough
to articulate what I was actually trying to build.

Not in corporate language.
Not in 20 pages of jargon.
Just simple prompts that force the thinking.

What is the problem.
Who feels the pain.
What success actually looks like.
What must exist now.
What can wait.

That tiny discipline unlocked a lot.

It made JournoPitch easier to steer.
It made AICRM easier to extend.
It made Replit easier to collaborate with.
It even made ChatGPT give me better answers
because I was finally giving it better questions.

The PRD Prompt Pack in today’s infographic
isn’t about documents.
It is about decisions.

A PRD is just a mirror that tells you
whether your idea has legs
or whether you are building your own trap.

And the moment you nail this,
your build speed goes up,
your revision loops shrink,
𝗮𝗻𝗱 𝘆𝗼𝘂𝗿 𝗳𝘂𝘁𝘂𝗿𝗲 𝘀𝗲𝗹𝗳 𝘀𝘁𝗼𝗽𝘀 𝘄𝗮𝗻𝘁𝗶𝗻𝗴 𝘁𝗼 𝗳𝗶𝗴𝗵𝘁 𝘆𝗼𝘂.

If you want to move quicker,
get better at thinking before typing.

Your prompts define your product.
Your clarity defines your velocity.

Nuff said.

Vibe Coding PRD Prompt Pack /in/zohemustafa

𝗛𝗼𝘄 𝗡𝗢𝗧 𝘁𝗼 𝘁𝗮𝗸𝗲 𝟭𝟴 𝘆𝗲𝗮𝗿𝘀
𝘁𝗼 𝗿𝗲𝗮𝗰𝗵 𝟭𝟬𝟬𝟬𝟬 𝗳𝗼𝗹𝗹𝗼𝘄𝗲𝗿𝘀.

As I perch on the precipice
of 10000 followers,
I’ve realised something.

𝘐 𝘮𝘢𝘥𝘦 𝘵𝘩𝘪𝘴 𝘢 𝘭𝘰𝘵 𝘩𝘢𝘳𝘥𝘦𝘳
𝘵𝘩𝘢𝘯 𝘪𝘵 𝘯𝘦𝘦𝘥𝘦𝘥 𝘵𝘰 𝘣𝘦.

And for context
my early LinkedIn era
was basically me treating it
like a very polite Facebook.
Random thoughts.
Random shares.
Zero intention.
Zero usefulness.
Peak chaos energy.

So here are the lessons
from someone who took
the scenic route.

𝟭. 𝗗𝗼𝗻’𝘁 𝗽𝗼𝘀𝘁 𝗿𝗮𝗻𝗱𝗼𝗺 𝘀#𝘁.*
Pick a lane.
Pick a niche.
Pick a topic.
People need to lock on
to your expertise.
The algorithm loves it too.
I find this boring
and I cannot stick
to one topic
if my life depended on it.
Since early 2024
I’ve posted about
AI tools,
marketing,
vibe coding,
AI adoption,
growth hacking,
automation,
avatars,
public speaking
and other completely
random things.
If no one knows
what I’m offering
that’s fine.
Neither do I lol.

𝟮. 𝗣𝗼𝘀𝘁 𝗱𝗮𝗶𝗹𝘆.
It matters.
Disappear for weeks
or months
and the algorithm
moves on without you.
Your audience does too.
Momentum is not loyal.
Miss your posting rhythm
and you’re basically
starting from zero again.

𝟯. 𝗗𝗼𝗻’𝘁 𝗰𝘂𝗹𝗹 𝘆𝗼𝘂𝗿 𝗰𝗼𝗻𝗻𝗲𝗰𝘁𝗶𝗼𝗻𝘀.
Yes it makes the network
more relevant.
But it absolutely
torches your numbers.
Also don’t be
a connection wh#*e
and accept every rando
who sneezes in your direction.
You’ll just end up
culling them later
and regretting both decisions.

𝟰. 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 𝗱𝗮𝗶𝗹𝘆.
Treat engagement
like it’s delaying
a sixth extinction event.
Seriously.
Comments compound.
Stop commenting
and you lose
the momentum you built.

5. 𝗜𝗳 𝘆𝗼𝘂’𝗿𝗲 𝘀𝗲𝗻𝗱𝗶𝗻𝗴
𝗰𝗼𝗻𝗻𝗲𝗰𝘁𝗶𝗼𝗻 𝗿𝗲𝗾𝘂𝗲𝘀𝘁𝘀
stop changing your offer
and your ICP every two weeks.
All you do is dilute your network
and create more culling
for Future You.
Future You hates this.

6. 𝗦𝘁𝗶𝗰𝗸 𝘁𝗼 𝗼𝗻𝗲 𝗵𝗲𝗮𝗱𝗹𝗶𝗻𝗲.
Don’t change it
17 gazillion times
like me.
Every time you switch it
you confuse your network.
And confused people
do nothing.

If I took 18 years
to reach 10000
you do not need to.

N.B. I am NOT 'Open To' 'Add Section'. 
That sounds like a really boring job 😜

How not to take 18 Years to reach 10000 followers /in/zohemustafa

When you’re Vibe Coding,
you don’t just build.
𝗬𝗼𝘂 𝘀𝗵𝗮𝗽𝗲𝘀𝗵𝗶𝗳𝘁.

There’s this myth that Vibe Coding
is mostly about learning tools
or writing perfect prompts.

It’s not.
It’s about thinking in roles.

When I built JournoPitch and AICRM,
I had to wear every hat in the room 
even the imaginary ones.

One hour I was a Business Analyst,
mapping how users might search for journalists
or score a lead in a CRM.

Next, a Designer,
sketching wireframes that made sense
to a first-time founder at 1am.

Then a Developer,
debugging JSON loops
that refused to behave.

Followed by Information Architect,
structuring everything
so it didn’t collapse under new features.

Then came Builder,
stringing it all together
in Replit, ChatGPT, V0 and Notion.

And when it was finally “done”
I became the Copywriter,
finding the words that explained it
to people who don’t yet know they needed it.

Then the Project Manager,
keeping myself accountable
without a team stand-up or Jira board.

The Product Lead,
deciding what not to build
(which is harder than it sounds).

And finally, the Marketer
and Growth Hacker,
testing how the world might discover
what I’d just created.

It’s messy.
It’s exhausting.
But it’s also the 𝗺𝗼𝘀𝘁 𝗳𝗿𝗲𝗲𝗶𝗻𝗴 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲
you can have as a creator.

Because you start to realise
you’re not just “learning tools.”
You’re learning to think
like an entire company.

That’s the 𝗺𝘂𝗹𝘁𝗶-𝗿𝗼𝗹𝗲 𝗯𝘂𝗶𝗹𝗱𝗲𝗿 𝗺𝗶𝗻𝗱𝘀𝗲𝘁.
You don’t wait for permission,
funding, or a team.
You just build
𝘸𝘪𝘵𝘩 𝘦𝘯𝘦𝘳𝘨𝘺, 𝘯𝘰𝘵 𝘴𝘺𝘯𝘵𝘢𝘹.

Question:
If you had to wear every hat
to build your next big idea,
which role would you struggle with most?

𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝗹𝗼𝘃𝗲𝘀 𝘀𝗵𝗼𝘄𝗶𝗻𝗴 𝗼𝗳𝗳 𝘁𝗵𝗲𝗶𝗿 𝗔𝗜 𝘀𝘁𝗮𝗰𝗸𝘀.
Mine fits in one screenshot.

Just Replit.
ChatGPT.
A sprinkle of V0 for mockups.
And time.

I didn’t use Make .com.
I didn’t design in Canva.
No fancy auto-builder dashboards.
I do use Make & Canva but for other stuff.

That’s it.
Three tools.
One mission: build real apps, not prototypes.

But what really powered JournoPitch and AICRM
wasn’t the tools.
It was how I used them.

Each Replit build started with a simple prompt:
“Help me build this…”
Then came the pattern:
Iterate. Test. Rethink. Re-prompt.

I never treated ChatGPT as a coder.
I treated it like a partner.
We debugged logic.
Redesigned flow.
Named functions like we were naming songs.

You don’t need a big toolchain.
You need dialogue.
And clarity of intent.

The rhythm between curiosity and clarity.
Between human instinct and machine precision.

If you can hold that rhythm,
the stack becomes irrelevant.
Because the build becomes inevitable.

What’s your smallest possible stack
that still gets big things done?

𝗬𝗼𝘂 𝗱𝗼𝗻’𝘁 𝗻𝗲𝗲𝗱 𝗮 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗽𝗹𝗮𝗻 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁.
You just need curiosity.

When I first started building JournoPitch, I didn’t have a full spec, design system, or roadmap.
Just a half-baked idea and too much caffeine.

But curiosity pulled me into Replit.
Two days later, I had a working beta.
Not perfect, not prettym, but alive.

That’s the magic of prototyping your curiosity.
You move from “what if” to “look at this.”
From thought to thing.

Most people wait for clarity before starting.
But clarity rarely comes before action.
Momentum creates understanding.

In those first 48 hours, I learned more
about what JournoPitch could be
than I ever could by writing about it.

Every feature you think you need
becomes irrelevant once the thing works.
Because then you can feel it.
You can see the gaps.
You can sense what’s missing.

And suddenly, curiosity becomes a compass.
Not just a spark.

If you’re sitting on an idea right now,
don’t wait until it’s fully formed.
Prototype your curiosity.
𝗕𝘂𝗶𝗹𝗱 𝘄𝗵𝗮𝘁 𝘆𝗼𝘂 𝘄𝗶𝘀𝗵 𝗲𝘅𝗶𝘀𝘁𝗲𝗱, 𝗲𝘃𝗲𝗻 𝗶𝗳 𝗶𝘁’𝘀 𝘀𝗰𝗿𝗮𝗽𝗽𝘆.
𝘌𝘴𝘱𝘦𝘤𝘪𝘢𝘭𝘭𝘺 𝘪𝘧 𝘪𝘵’𝘴 𝘴𝘤𝘳𝘢𝘱𝘱𝘺.

𝗔𝗽𝗽𝗮𝗿𝗲𝗻𝘁𝗹𝘆, 𝗩𝗶𝗯𝗲 𝗖𝗼𝗱𝗶𝗻𝗴 𝗶𝘀 𝗱𝗲𝗮𝗱.
𝗔𝗴𝗮𝗶𝗻. Or never was a solution.

Every few months, someone declares that Vibe Coding has imploded.
That it’s all “duct-taped disasters”, hallucinated libraries,
and wannabe founders typing dreams into AI models.

And I get it.
If you treat Vibe Coding like a toy, it will behave like one.
But if you treat it like a system, it 𝗰𝗮𝗻 𝗯𝘂𝗶𝗹𝗱 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻-𝗴𝗿𝗮𝗱𝗲 software.

JournoPitch and AICRM weren’t born out of hype.
They were built line by line, prompt by prompt, debug by debug 
using the same so-called “vibe tools” people love to ridicule.

What’s funny is the critics keep describing a problem
that doesn’t exist for those who build with intent.
Security? Replit handles API isolation beautifully.
Scalability? You design for it.
Structure? That’s what PRDs are for.

The issue isn’t the tools. It’s the mindset.

And here’s something I’ve noticed.
Most of the people shouting “fad!”
are either old-skool AI builders [aka '𝘱𝘳𝘰𝘱𝘦𝘳 𝘥𝘦𝘷𝘴']
or founders of dev agencies whose business model depends on gatekeeping.
They mock what threatens the moat they built around complexity.

No irony lost there.

I’ve had quotes north of £12,000 just for an MVP.
Another “expert” said AICRM would need £500k - £1M to build.
If I’d listened, neither JournoPitch nor AICRM would exist today.

So when people say Vibe Coding is a fad, I smile.
Because while they’re writing think-pieces about why it can’t work,
I’m shipping updates.

You wants a quiet truth no one wants to admit?:

Vibe Coding doesn’t eliminate engineering.
It redefines it.
You still need logic, design sense, empathy for users,
but you also need imagination,
the courage to iterate fast,
and the humility to ask the right prompts.

We’ve moved from “write code”
to “design conversations that build code”.

That’s not regression to moi.
𝘛𝘩𝘢𝘵’𝘴 𝘦𝘷𝘰𝘭𝘶𝘵𝘪𝘰𝘯, 𝘣𝘶𝘣.

So if you’re building in public,
don’t let the cynicism slow you down.
You’re not “vibe coding”.
You’re re-engineering what it means to create.

Have you noticed how every tech revolution starts as a joke before it becomes the standard?

𝗔𝗴𝗲𝗻𝘁 𝗙𝗶𝗿𝘀𝘁 𝗜𝗗𝗘𝘀. 𝗔𝗻𝗱 𝘄𝗵𝗮𝘁 𝗚𝗼𝗼𝗴𝗹𝗲 𝗔𝗻𝘁𝗶𝗚𝗿𝗮𝘃𝗶𝘁𝘆 𝗺𝗶𝗴𝗵𝘁 𝗺𝗲𝗮𝗻 𝗳𝗼𝗿 𝗯𝘂𝗶𝗹𝗱𝗲𝗿𝘀.

I keep seeing posts about Google’s new AntiGravity IDE.
Everyone is breathless.
The future has apparently arrived.

𝘐 𝘩𝘢𝘷𝘦𝘯’𝘵 𝘦𝘷𝘦𝘯 𝘶𝘴𝘦𝘥 𝘪𝘵 𝘺𝘦𝘵. 😆

But I am watching the shift it represents.

For the last year I’ve been building JournoPitch
and AICRM with Replit as my home base.
Vibe Coding in its purest form.

And even then… it takes time.
Energy.
Context juggling.
Feature creep.
Roadmaps that mutate while you sleep.

So when a platform like AntiGravity appears,
the promise is seductive.

A fully agent-first IDE.
Code, refactor, test, ship.
All in one continuous motion.
It reads your intent, not just your instructions.

The interesting part isn’t the shiny UI.
It’s the pattern.

Replit was first to genuinely make
AI-native building feel real.
Claude Code brought the deep reasoning layer.
Lovable leaned into full-flow scaffolding.

AntiGravity is pushing the 𝗢𝘃𝗲𝗿𝘁𝗼𝗻 𝘄𝗶𝗻𝗱𝗼𝘄 again.
Less coding.
More orchestration.
More “tell the system what outcome you want
and let it negotiate the rest.”

Even if AntiGravity is astonishing
I don’t know how I would migrate a live product.
JournoPitch has a credit engine,
Stripe logic,
monitoring pipelines,
editor flows,
and a roadmap stacked with new features.

AICRM is the same.
Deep wiring everywhere.

Does AntiGravity deploy?
Does it manage infra?
Does it handle persistent agents?
Is there a safe bridge from what exists
to what might exist?

And more importantly
𝗱𝗼 𝘄𝗲 𝘂𝗻𝗱𝗲𝗿𝗲𝘀𝘁𝗶𝗺𝗮𝘁𝗲 𝘁𝗵𝗲 𝘀𝘄𝗶𝘁𝗰𝗵𝗶𝗻𝗴 𝘁𝗮𝘅
because the marketing hype is louder
than the migration reality?

I love new tools.
But I also love finishing what I start.

Maybe the future isn’t choosing one IDE.
Maybe it’s learning how to think
in an agent-first way
no matter where you build.

Agent First IDEs

𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝘁𝗮𝗹𝗸𝘀 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝘁𝗼𝗼𝗹𝘀.
Almost no one talks about 𝘵𝘩𝘦 𝘴𝘺𝘴𝘵𝘦𝘮.

That’s why most “no-code” projects stall at the prototype.

Because tools don’t build production-grade apps, 𝗽𝗿𝗼𝗰𝗲𝘀𝘀 𝗱𝗼𝗲𝘀.

Vibe Coding isn’t about syntax.
To me it’s about energy.

It’s creative building powered by AI, intuition, and systems thinking.
Where ideas become working apps; live, public, playful.

Here’s the system I use:

1. Challenge → What’s broken or interesting?
2. Context → Why now? What already exists?
3. Outcome → What does ‘done’ look like?
4. Subtasks → Break it into micro-wins.
5. Feedback → Test, share, iterate.

This loop built @JournoPitch and @AICRM.
It’s the same rhythm I use every time.

AI supports the logic, debugging, and ideas
ChatGPT, Gemini, Claude.

Replit takes it from draft to deployment,
secure and production-ready.

Not a showcase of tools.
A repeatable process that stacks curiosity, clarity, and iteration.

If you can explain it, you can build it.

What is Vibe Coding, A Guide to building with Energy not Syntax /in/zohemustafa

𝗪𝗵𝗲𝗻 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝗶𝘀 𝗼𝗽𝘁𝗶𝗺𝗶𝘀𝗲𝗱,
𝘄𝗵𝗮𝘁 𝘀𝘁𝗶𝗹𝗹 𝗳𝗲𝗲𝗹𝘀 𝗵𝘂𝗺𝗮𝗻?

This week’s Machine Minds dives into the
shift no marketer can ignore.

AI is now the baseline.
Agentic workflows run campaigns.
Creators outperform traditional media.
Attention is the new currency.

But the brands winning right now?
They’re doubling down on something that
automation can’t simulate.

Tone.
Trust.
Emotional clarity.
Human resonance.

𝘐𝘧 𝘦𝘷𝘦𝘳𝘺 𝘴𝘺𝘴𝘵𝘦𝘮 𝘣𝘦𝘤𝘰𝘮𝘦𝘴 𝘱𝘦𝘳𝘧𝘦𝘤𝘵,
𝘺𝘰𝘶𝘳 𝘩𝘶𝘮𝘢𝘯𝘪𝘵𝘺 𝘣𝘦𝘤𝘰𝘮𝘦𝘴 𝘵𝘩𝘦 𝘴𝘵𝘳𝘢𝘵𝘦𝘨𝘺.

Digital marketing just crossed a threshold. Every workflow is becoming automated. Every decision is being predicted. Every channel is optimised by def

Open article: When Everything Is Optimised, What Still Feels Human? by Zohe Mustafa • 4 min read

When Everything Is Optimised, What Still Feels Human?

𝗜’𝘃𝗲 𝗱𝗲𝗰𝗶𝗱𝗲𝗱 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗶𝗻 𝗽𝘂𝗯𝗹𝗶𝗰.
Not to show off. [𝘛𝘩𝘢𝘵'𝘴 𝘳𝘦𝘢𝘭𝘭𝘺 𝘯𝘰𝘵 𝘮𝘦 😱]
But to show up.

For years, I've built quietly.
Behind closed tabs, half-finished prototypes, and ideas that never saw daylight.
Because silence feels safe.
Until you realise no one can see what you’re actually building.

So this time, I’m doing it differently.
I’m building out loud. Well, not 𝘁𝗼𝗼 loud.
Sharing the messy middle, the progress, and the lessons as they happen.

When you build in public, you stop chasing polish and start chasing momentum. [Thats the plan anyway].
You stop hiding your learning curve and start turning it into your superpower.
You replace “I’ll share when it’s done”
with “Here’s what I learned today.”

It’s not marketing.
It’s accountability.
It’s energy.
It’s curiosity.

Whether it’s JournoPitch, AICRM, or something completely new,
I’m building in public because it keeps me honest,
focused, and hopefully connected to others on the same path.

AICRM I've been trying to get built since 2022.
Many, many people said it would not be possible to build it.
𝘚𝘰𝘮𝘦 𝘦𝘷𝘦𝘯 𝘴𝘢𝘪𝘥 𝘭𝘦𝘢𝘷𝘦 𝘪𝘵 𝘵𝘰 𝘵𝘩𝘦 𝘣𝘪𝘨 𝘣𝘰𝘺𝘴 𝘭𝘪𝘬𝘦 𝘏𝘶𝘣𝘚𝘱𝘰𝘵 & 𝘚𝘢𝘭𝘦𝘴𝘍𝘰𝘳𝘤𝘦
I had quotes of £𝟭𝟮.𝟱𝗞 𝗷𝘂𝘀𝘁 𝗳𝗼𝗿 𝗮𝗻 𝗠𝗩𝗣.
£𝟱𝟬𝟬𝗞 𝘁𝗼 £𝟭𝗠 to build the full app.

I now have a more than a MVP for a tiny fraction of the quoted costs..
Its buggy, needs a ton more testing..... but I'm getting closer.

If you’ve been sitting on something you want to create,
this is your sign.
Start small.
Share the thinking before the product.
That’s where the real momentum begins. 🤞

Why you should build in public. Even more so if its vibe coded and the compounding rate of AI advancements. Follow /in/zohemustafa

𝗠𝗼𝘀𝘁 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝗮𝗱𝘃𝗶𝗰𝗲 𝗳𝗼𝗰𝘂𝘀𝗲𝘀 𝗼𝗻 𝗳𝗲𝗮𝘁𝘂𝗿𝗲𝘀.

𝗔𝗹𝗺𝗼𝘀𝘁 𝗻𝗼𝗻𝗲 𝗳𝗼𝗰𝘂𝘀𝗲𝘀 𝗼𝗻 𝗳𝗹𝗼𝘄.

And yet flow is what decides whether someone
uses your product once
or returns tomorrow.

I learnt this the slow way.

Every time I built something 
JournoPitch, AICRM, or even the early “rough” things
the outcome wasn’t shaped by which feature I added next.
It was shaped by the path the user walked through.

If the path was clean
the product felt smart.

If the path was messy
the product felt messy
even if the features were strong.

That’s when I realised something important.
You can ship a product that is technically impressive
and still lose users
because the journey feels confusing.

𝘠𝘰𝘶𝘳 𝘶𝘴𝘦𝘳 𝘪𝘴 𝘯𝘰𝘵 𝘪𝘯𝘴𝘪𝘥𝘦 𝘺𝘰𝘶𝘳 𝘩𝘦𝘢𝘥.
They only see the steps.

So today’s infographic is a collection of frameworks
I wish I’d had years ago.
Not screens.
Not layouts.
Not UI kits.

Thinking tools.

Stuff that makes you a better builder,
not just a faster one.

Principles that stop you guessing
and force you to design journeys that actually convert.

The frameworks inside the graphic are the ones I rely on
They help you compress, simplify, interrogate
and pressure-test your flow
so the UI becomes the expression of good thinking
not a patch for bad structure.

Flow Frameworks

Fahd Baig

Capgemini

Senior Data Scientist / AI Engineer

QoqoFactory.ai

QoqoFactory.ai is consultancy focussed on solving business and customer issues using bleeding-edge AI and ML advancements such as LLMs.

Currently developing use cases in Customer Services, Marketing & Campaigns and Knowledgebase & Innovation for businesses as a service. 

Also developing Prompt Design and Prompt Engineering guidance and best practice scripts for relevant models.

Using Open LLMs so businesses can control the data being used to tune the models and add a governance layer over the LLM models.
 
Please get in touch if any of these feel relevant and you would need expert assistance for your business.

Lead AI Consultant

esynergy

ESynergy is a technology consultancy that is laser focussed on providing measurable business value to its clients. Mainly working within highly regulated industries, with organisations of all sizes from large enterprises to scale-ups, building scalable platforms, products, and services.

I focus on bringing AI, Data Science and Machine Learning expertise to our clients and taking them on a journey to realise significant value from data.

Lead Associate - AI, Data Science and Machine Learning

Cabot Credit Management

Cabot is the largest credit management service provider in the UK and Europe's largest debt purchaser with a customer-centric business model and large customer-specific datasets

As a Lead Data Scientist, I am responsible for using customer data to develop models that bring customers' perspective into company's decision making. I lead the Advanced Analytics team within Data & Innovation and enshrine the 'Full Stack' ML approach from managing stakeholders, EDA and, developing ML models to deploying/productionising models and, monitoring and feedback. I am also involved with Data Strategy and selling the 'Art of the Possible with ML' to the leadership across Cabot. The team focuses on 1) Operations Optimisation 2) Digital 3) NLP

I have been involved in developing many production grade ML models deployed on-prem & on Azure Databricks:

1) Led a paradigm shift in business strategy by changing generic strategies to become customer-centric. Developed Customer Classification model using unsupervised modelling (KMeans/GMMs) & rich socio-demographic customer data (OpenAPI) to develop homogenous groups of customers (over 4m customers) using tSNE/UMAP visualisations & quantitative analysis to define segments. Developed using Python, microservices, SQL, Scikit-Learn on Databricks MLflow to deploy models
2) Improved accuracy of predicting customers that own properties outright to improve performance of litigation team. Worked with highly imbalanced datasets and hypothesis based experimentation to develop an optimised model. 
3) Improved productivity by developing NLP based model to extract information from customers within thousands of agent notes to automatically select customers for specific treatment. This not only lead to cost-savings but laid foundation  within Cabot for future NLP projects. Used N-grams/BERT vectors with SparkML and worked with large set of past notes containing noisy labels, providing pragmatic solutions to develop quality model within the deadline

Lead Data Scientist - Machine Learning

Cabot Financial

Money Dashboard

Money Dashboard is UK’s largest, much loved, award winning personal finance app that helps its users understand and manage their finances. Money Dashboard is on a mission to make people happier and more successful by putting them in control of their finances. It is fixated with its users and puts data at the heart of everything it does to develop incredibly intuitive, highly innovative products. 

As Head of Data Science, I am responsible for establishing a culture of data-driven decision making within Money Dashboard. This entails developing processes and insights that help different teams (Data, Tech, Product, Support and Senior Management) access, utilise and benefit from insights within data. I am also involved in client and key internal stakeholder communication reporting on progress, information and potential bottlenecks to key Data Science & Machine Learning projects.

I have been involved in developing a production grade machine learning categorisation engine that categorises the purpose and merchant for user's bank transactions. This is a fundamental building block in developing innovative products within FinTech and therefore I have been spearheading efforts to rollout Categorisation As A Service to the wider FinTech community:
-Developed using out of core supervised learning Stochastic Gradient Descent based classifier. 
-Trained models on over 400 million transactions obtained from over 200k user, 450k accounts and 60 financial institutions.
-Classify transactions across 185 purpose categories (over 3 level of granularity) and 1200 Merchants (more added every month).
-Leverage Money Dashboard's deep expertise and domain knowledge in developing intricate, complex set of heuristics to develop models.
-Develop processes to train, test and deploy models into production.
-Develop micro-service based highly responsive and scalable system on AWS using AWS SQS, S3 and Elasticache.
-Developed Machine Learning using Python and Scikit ML libraries.
-Agile,Jira,Git

Head of Data Science

UNIVERSITY OF DUNDEE(THE)

Machine Learning and Data Mining
-Classification systems: Support Vector Machines (SVMs), Neural Networks, Random Forests, Bayesian Classifiers, HMMs
-Regression based systems for forecasting and predictive analytics: Linear, Logistic, SVM and Neural Network based regression
-Clustering: K-means, Agglomerative clustering
-Deep Learning algorithms and Convolutional Neural Nets for computer vision and speech recognition application
-Dimensionality Reduction: LDA, PCA
-Applications: Handwriting/OCR system, Speech based systems, Natural language processing (NLP)
-Tools: Azure Machine Learning, AWS MAchine learning, Python, R, MATLAB

Big Data Technologies
-Lambda Architecture, Data pipeline: Hadoop, Spark, Couchbase, Hive, and, Pig and Erlang

Business Intelligence and Analysis
-Business Requirement along with Sun and STAR modelling analysis
-ETL and Data Warehouse: SQL queries, SQL Server (2014/2016), SSAS, SSIS
-Data visualisation and BI tools: Power BI, Tableau, D3 

Technology and Innovation Management
-Product management with lean start-up and Agile methodologies
-Mawenzi: Intelligent product prioritisation platform to help Product Managers brainstorm concepts, collate ideas, prioritise features and measure outcomes

MSc Project placement - Money Dashboard
-Machine Learning based system that identifies the purpose an individual's financial transaction
-Used online supervised classification system using Stochastic Gradient Descent based logistic regression models (mini-batch training)
-Trained models on a large dataset (in excess of 300 million transactions) across 289 categories and 3 level of granularity

MSc Data Engineering

G2G3 (part of Capita plc)

G2G3 is an innovative technology company that develops simulations, serious games and gamification products for IT enterprises to enable engagement and drive positive behaviour within staff, partners or clients.

Technical Manager

FB Consulting

FB Consulting works with innovative technology companies to develop their market/product and assist them in raising finance. The aim is to provide companies with key management skills to develop their businesses further.

Director Management Consulting

QoqoFactory

Focused on developing the next generation of mobile apps and games for Android, iPhone and Windows platform. Leading the development of mobile apps that will focus on HTML5 and games with high-end graphics particularly nVidia's CUDA technology and Tegra4 platform.

Founder

I am an experienced, entrepreneurially-driven data leader with a global mindset. I find analysing data and uncovering patterns and relationships hidden within it that are not apparent otherwise, really brings data to life for me. The ability to then take these insights and communicate it as a coherent, engaging story to management and clients thereby enabling data-driven decision making, strategic planning and data-supported development of products and business models really excites me. I am passionate about what data can do and believe it can allow an organisation to leap-frog its competition, find new markets and develop solutions to hard problems all through actionable insights from data. I have developed a number of production grade machine learning products and am looking for companies who are solving complex problems using data. I am interested in developing data systems and pipelines that allow businesses to leverage their data assets. I aspire to work with innovative companies and help them gain competitive advantage, enter new markets and solve complex challenges using data.

I have been awarded the prestigious SFC funding to complete MSc in Data Engineering from Dundee University. I have also graduated from a highly prestigious and competitive Saltire Fellowship Programme at Babson College, USA (Ranked #1 for MBA in Entrepreneurship for 28 consecutive years). This is a fully-sponsored, accelerated experiential MBA style programme that aims to enhance global entrepreneurial leadership skills.

In the past I have successfully led an award-winning technology business, in setting its technological direction, hiring key personnel, building corporate structure and raising funding. My key strengths include creative thinking, tenacity to surmount business challenges, inspire people and motivate them to perform their best. 

I aim to harness my entrepreneurial, leadership and data skills & experiences to lead new innovative technological ideas or departments, within large organisations, to global success.

taplio

The Battle

Zohe Mustafa

Fahd Baig

Followers

Posts

Engagements

Likes

Performance ScoreZohe Mustafa's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Zohe Mustafa

Fahd Baig

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Zohe Mustafa's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreZohe Mustafa's LinkedIn Performance