I create content about deep learning | The DevRel and Deep Learning Dude

Hacking on LLMs, curating multimodal datasets, and writing a book about RAG.

Harpreet Sahota 🥑

i've literally been waiting for years to have an OPEN SOURCE model like qwen3-vl-embedding

i tested its multimodal retrieval capabilities on a corpus of 412 short video clips and the results literally blew my mind

here are the queries i tested:

1. a cartoon guy drinks merlot wine

i like this query because we see how it can retrieve based on semantics (a cartoon), text (the label merlot), and temporal action in the video (the cartoon guy drinks the wine mid-way through the video)

2. garfield runs out of the door

semantics are awesome here, the model knows i'm talking about the cartoon character...

3. a woman in a blue dress sleeping on a red bench

the skeptic in you might think that it's retrieving just based on the red color block in the 2/3 of the video...but notice the specific part of the query "a woman in a blue dress..." and this is only shown in 3s out of the full 10s

this is such a huge release 

and it's gonna open up SO much more for multimodal video retrieval this year

on my wishlist is natural language search of pcd datasets, who gonna ship that?

you can hack around with the model using the resources below

check out the docs: here: https://lnkd.in/gh7yfBNP

and the quickstart nbhere:  https://lnkd.in/gbinsx7p

🤖 Hacker-in-Residence @ Voxel51| 👨🏽‍💻 AI/ML Engineer | 👷🏽‍♀️ Technical Developer Advocate | Learn. Do. Write. Teach. Repeat.

for those who missed it, or were there in-person and know the trouble i was facing 😆, my ODSC webinar is now live on youtube!

in this session i discuss visual ai techniques for automotive damage detection. 

i used the cardd dataset, which is a public dataset built specifically for vehicle damage analysis and demonstrated how fiftyone helps you explore, visualize, and build models to detect dents, scratches, cracks, glass shatter, tire flats, and broken lamps.

i feel like its a super practical workshop with some insightful workflows for visual AI in automotive settings, but as the creator i might be biased.

only one way to find out for yourself, watch the webinar!

https://lnkd.in/ewfz5vZD

Open video: Mastering Visual AI with Vision-Language Models & Advanced Evaluation Techniques by Harpreet Sahota on youtube.com

Mastering Visual AI with Vision-Language Models & Advanced Evaluation Techniques by Harpreet Sahota

sometimes i look back at the shit i get done and i genuinely impress myself

how teams really build Visual AI in the wild? what's the he messy reality?  

i don't know...and maybe its because nobody’s asking practitioners and compiling their answers so we can find out the landscape actually looks like. 

so, we at Voxel51 are kickin off the...

wait for it...

 State of Visual AI survey...vey...vey

for those out ther building visual AI systems i want your hot takes

what’s the hardest part about making visual ai a reality? 

let us know in this survey

it'll take you 30 minutes and is a much better use of time than baby sitting your runs and staring at your loss curves

i can't offer much beyond my heartiest thanks, but you will get early access to the full report with benchmarks, common pain points, and where you stand within the landscape. 

here's a link: https://lnkd.in/gnsADACy

MedGemma 1.5 supports detection, but for best results, you'll need to fine-tune. 

there's a kaggle competition using the model, i created a starter notebook to give you a jump start on how to fine-tune it for detection

checkout the starter notebook and additonal resources 👇🏼

cc Omar Sanseviero Megan Risdal Logan Kilpatrick

looking forward to the next meetup in boston!

Join our in-person meetup at the Microsoft Research Lab at the Massachusetts Institute of Technology in Boston, Massachusetts on Feb 18 to hear talks from experts on cutting-edge topics across AI, ML, and computer vision. Register here: https://lnkd.in/gi4kpSMB

Talks will include:

• SNAP: Towards Segmenting Anything in Any Point Cloud - Hanhui Wang at Northeastern University

• Culturally Adaptive AI - Anku R. at Massachusetts Institute of Technology

• Data Foundations for Vision-Language-Action Models - Harpreet Sahota 🥑 at Voxel51

• Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models - Sharut G. at Massachusetts Institute of Technology

• Neural Radiance Fields for Image Verification - Sage Simhon at Pasteur Labs & ISI

#computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

Syed Abdul Sami Rizvi, Matthew Ebisu and 118 other attendees

View event: Feb 18 - Boston AI, ML and Computer Vision Meetup

Feb 18 - Boston AI, ML and Computer Vision Meetup

in the asu area and wanna get started with open source contributions? then don't miss this hack

On March 21, 2026, we’re hosting Agents World: Visual AI Hackathon @Arizona State University, together with Pavan Turaga 🚀 

This is a hands-on day: meet people, build with AI, eat pizza, and leave with something you created after a short workshop and a full day of hacking. 
You’ll work with the FiftyOne ecosystem and real models used in practice, like Qwen, Gemini Vision, Sharp, and more. 

If you’re an ASU student who wants to build, learn by doing, and work with real Visual AI tools, this is for you.

Registration is open:
https://lnkd.in/gBCHW5Fg

Hope to see you at ASU. I’ll be there with the amazing Voxel51 team:  Sid Mehta, Ritchie Martori, Jimmy Guerrero ! 🙌

Inspired by the amazing Hugging Face’s 2025 AI Timeline, we created one for the Fiftyone community. The FiftyOne Community Development Timeline highlights more than 131 contributions from 2025, including datasets, models, plugins, and integrations.

You can filter by type or modality and jump straight to each integration.
https://lnkd.in/ev2kpCGN

Big thanks to the entire Voxel51 community for building all of this together.
Harpreet Sahota 🥑, Daniel Gural, Paula Ramos, PhD, Jimmy Guerrero, Derek Van de Streek, Nicholas Lotz 🚀

Gonna be a great series of talks, lots of technical depth!

Chief Community Officer @ Voxel51 | Startup Advisor

Join us on Feb 18 for the Boston AI, ML and Computer Vision Meetup! Pre-registration is mandatory to clear building security - https://lnkd.in/gHtBuH49

Talks will include:

* SNAP: Towards Segmenting Anything in Any Point Cloud - Hanhui Wang at Northeastern University

* Culturally Adaptive AI - Anku R. at MIT

* Data Foundations for Vision-Language-Action Models - Harpreet Sahota 🥑 at Voxel51

* Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models - Sharut G. at MIT

* Neural Radiance Fields for Image Verification - Sage Simhon at Pasteur Labs & ISI

***********

Want to build better computer vision models? FiftyOne is an open source toolkit from Voxel51 (our Meetup sponsor) that helps you curate datasets, evaluate model performance, visualize embeddings, catch annotation errors, and eliminate duplicate images—all in one place.

“pip install fiftyone” is all it takes to get started - https://docs.voxel51.com/

#computervision #ai #artificialintelligence #machinevision
#machinelearning #datascience #physicalai

Open to Collaborate | Storyteller | Lead Data Engineer@Wavicle| Linkedin Top Voice 2024 | Globant | Linkedin Learning Instructor | 2xGCP & AWS Certified | LICAP'2022

Data Engineer with 7+ years of experience working on data engineering, analysis, and modelling projects for various domains, such as banking, healthcare, sales, and e-commerce. I leverage my strong statistical and analytical ability to explore business ideas and provide innovative solutions. 

Skilled with technologies and strong intellect with enthusiasm as an avid learner with challenging opportunities. I’m working as a LinkedIn Learning Instructor to bring change as a data evangelist.

Post Graduated in Master of Technology in Data Science and Engineering from Birla Institute of Technology and Science, Pilani in 2023, where I learned advanced concepts and techniques in data science and engineering, machine learning, cloud computing, and visualization. 

I hold multiple certifications in GCP, AWS, and Azure, and I enjoy working with cloud technologies and big data platforms.

I am passionate about data science and women empowerment, and I have been recognized as the 
- Linkedin Top Voice 2025, 2024
- Data Science and Women Empowerment Thought Leader by Global AI Hub. 

I am also a proud participant of the LinkedIn Creator Accelerator Program 2022, where I share my insights and experiences with the data science community. 

I am open to collaborate and share ideas, or explore various opportunities in the data field.

Pooja Jain

Learn, Earn or Leave

Loyalty is beautiful… until you realize  
it was never mutual.

Your company will replace you faster than you can say "team player"

The company?  
It’s just business.  
Treat it accordingly. 🛡️

Agree?

No alternative text description for this image

Data pipelines are evolving from the "copy-paste" era to the "access-anywhere" era. Are you evolving with them?

At its core, data pipelines answers one thing -
How does data moves from "happened" to "decision made"?

Here's how the evolution looks like:
𝗘𝗧𝗟 (The Old Way) Extract → Transform → Load
 • Like preprocessing your groceries in the parking lot before bringing them inside
 • Great for strict governance, legacy systems
 • Slow. Rigid. Breaks on every schema change
 • Tools: Glue, Talend, NiFi

𝗘𝗟𝗧 (The Cloud Era) Load raw → Transform in warehouse
 • Load raw, transform in warehouse. Like, Dump groceries in fridge, prep later
 • Scales with cloud compute
 • Better, but still copying data everywhere
 • Tools: dbt, Snowflake, BigQuery

𝗖𝗗𝗖 (Real-Time Tax) Change Data Capture
 • Stream every database change via transaction logs
 • Captures inserts/updates/deletes from source. Near real‑time, low overhead
 • Perfect for fraud detection. Expensive for everything else
 • Tools: Debezium, AWS DMS

𝗦𝘁𝗿𝗲𝗮𝗺𝗶𝗻𝗴 (Over-Engineering Champion) 
 • Real‑time processing, no waiting. Just like Food truck serving as you arrive
 • Necessary for IoT, trading, Fraud and live dashboards
 • Overkill for your Monday morning dashboard
 • Tools: Kafka, Flink, Kinesis.

Enter 𝗭𝗲𝗿𝗼 𝗘𝗧𝗟: The "Why Are We Doing This?" Moment
The modern lakehouse game‑changer
One city library for everyone – no need to photocopy books.
Data lives in one place (lakehouse).
Query it directly with any engine. No copying.

🤔 How Modern Lakehouses Make This Real?

𝗢𝗽𝗲𝗻 𝗧𝗮𝗯𝗹𝗲 𝗙𝗼𝗿𝗺𝗮𝘁𝘀 (Iceberg, Hudi, Delta)
Think: Git for data. ACID transactions, time travel, schema evolution—on cheap object storage.

𝗦𝘁𝗼𝗿𝗮𝗴𝗲 ≠ 𝗖𝗼𝗺𝗽𝘂𝘁𝗲
Data sits in S3. Spark reads it. Snowflake queries it. Trino analyzes it. Same files. No copies. No vendor lock-in.

𝗡𝗮𝘁𝗶𝘃𝗲 𝗖𝗼𝗻𝗻𝗲𝗰𝘁𝗼𝗿𝘀
AWS Aurora → S3? Managed replication.
No custom Glue jobs. No 3 AM debugging.

What Changes for You?

You stop:
- Writing extraction code for the 47th time
- Maintaining Airflow DAGs that "mysteriously fail on Wednesdays"
- Explaining why reports are 6 hours delayed
- Paying for storage in four different systems

You start:
- Designing smart partitioning strategies
- Building dbt models where transformation actually matters
- Focusing on data quality, not data plumbing
- Solving business problems instead of pipeline problems

The Real Talk?

Zero ETL isn't "no data engineering." 
It's engineering that matters.
Traditional ETL won't vanish overnight—too much legacy infrastructure. 

But for new projects? Start lakehouse-first:
- Store once in open formats
- Let compute engines come to data
- Create copies only when necessary

Think of it this way: 
You wouldn't photocopy a book every time someone wants to read a different chapter. 
So why copy your data every time a new team needs access?

Without governance, your data is a 10-lane highway with no traffic lights.

Everyone's driving. Nobody knows the rules. Accidents are inevitable.

Why does it Matters?
Think traffic system for organizational data.
 • Data flows smoothly, safely, smartly
 • Everyone knows their lane
 • Decisions made with confidence
 • Teams drive results, not fix messes

Without traffic lights, you don't get speed; you get gridlock and crashes.

🔑 Your Role as a Leader
Strategy & Oversight:
 → Champion governance (sponsorship + resources)
 → Set the tone for responsible data use
 → Hold leaders accountable for data quality
 → Not your job: Managing day-to-day data operations

👥 Who Does What ?(Simplifying the roles)
 • Chief Data Officer: Sets data strategy
 • Data Security Officer: Shields from risk
 • Head of Governance: Enforces data rules
 • Data Quality Manager: Validates accuracy

 → Data Owners: Accountable for specific datasets
 → Data Stewards: Maintain quality and access
 → Product Owners: Align data with user needs
 → Data Custodians: Handle storage and security

Here's to quick start and win early -
Reflect: Where do you rely on data today? Where do you trust it?
Clarify: Which outcomes matter most?
Identify: Who owns critical data decisions?

But, Why Now?
AI is scaling. Without governance, pilots stall and ROI evaporates.
Competition moves fast. Trusted data = speed advantage.
Regulations tighten. Fines and reputation hits are expensive.

𝘎𝘰𝘰𝘥 𝘨𝘰𝘷𝘦𝘳𝘯𝘢𝘯𝘤𝘦 𝘪𝘴𝘯'𝘵 𝘣𝘶𝘳𝘦𝘢𝘶𝘤𝘳𝘢𝘤𝘺—𝘪𝘵'𝘴 𝘵𝘩𝘦 𝘣𝘭𝘶𝘦𝘱𝘳𝘪𝘯𝘵 𝘧𝘰𝘳 𝘈𝘐 𝘵𝘩𝘢𝘵 𝘢𝘤𝘵𝘶𝘢𝘭𝘭𝘺 𝘥𝘦𝘭𝘪𝘷𝘦𝘳𝘴.

My final thoughts:
If 𝗱𝗮𝘁𝗮 is the fuel and 𝗔𝗜 is the engine,
𝗚𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 is the steering wheel.
Without it, speed means nothing.

Refer this to explore what Atlan says about Data Governance pillars in 2026 -https://lnkd.in/dAMTYJGy

Data Mesh for Data Engineers!
Most organizations still treat data like a giant centralized kitchen — slow, overloaded, and always backlogged.

Data Mesh flips that model.
Instead of one giant data team owning everything...
 → Domains own their data
 → Data is treated like a product
 → Platform teams provide self-serve infrastructure
 → Governance is federated, not chaotic

Think of multiple cafés, each owning its menu, ingredients, and customers 
That’s exactly how Data Mesh works
Decentralized, domain-owned, product-like data.

𝗪𝗵𝗮𝘁 𝗱𝗼𝗲𝘀 𝘁𝗵𝗮𝘁 𝗺𝗲𝗮𝗻?
Clear data contracts (no surprises downstream)
Built-in quality checks (catch issues before users do)
Versioned interfaces (breaking changes = major versions)
Metadata that actually helps people find your datasets

𝗗𝗼𝗲𝘀 𝘁𝗵𝗲 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗟𝗮𝘆𝗲𝗿 𝗠𝗮𝘁𝘁𝗲𝗿𝘀?
A mesh only works if engineers get:
Self-serve compute & storage
Standardized tooling (catalog, lineage, observability)
Guardrails, not gatekeepers
Without this? “Decentralization” just becomes distributed chaos.

𝗧𝗲𝗰𝗵 𝗶𝘀𝗻'𝘁 𝘁𝗵𝗲 𝗛𝗮𝗿𝗱 𝗽𝗮𝗿𝘁?
Data Mesh is an operating model shift:
Ownership moves to domains
Accountability follows ownership
Governance becomes federated and automated

It's less about new tools and more about changing how data responsibility works.
Done right, Data Mesh turns data engineering from a service desk into product engineering for data.

⚠️ Not a silver bullet — it needs strong governance + org-level change.
But when done right?
👉 It unlocks speed, scalability, and true data ownership.

💬Have you worked with Data Mesh? What actually worked vs what was just theory?

I watched a junior engineer spend 6 months building a Kafka-Spark-Kubernetes pipeline.
It crashed on day one in production.

Why? They never learned how to handle CSV files properly.

Here's what nobody tells you about data engineering:

The staircase exists for a reason. 
𝐒𝐤𝐢𝐩 𝐬𝐭𝐞𝐩𝐬, 𝐚𝐧𝐝 𝐲𝐨𝐮 𝐝𝐨𝐧’𝐭 𝐦𝐨𝐯𝐞 𝐟𝐚𝐬𝐭𝐞𝐫—𝐲𝐨𝐮 𝐟𝐚𝐥𝐥 𝐡𝐚𝐫𝐝𝐞𝐫.
You can't skip fundamentals without consequences.

Here’s the order that actually holds up in the real world:
  • SQL → then Snowflake
 • Python → then Databricks
 • Warehouse basics → then dbt
 • Files and formats → then data lakes
 • Relational databases → then NoSQL
 • Batch jobs → then streaming
 • Partitions → then performance work
 • Linux and networking → then cloud
 • Business context → then tuning queries
I’ve seen this play out again and again.

Two paths. Same destination:
Path 1: Climb every step—Databases, SQL, Python, ETL, Snowflake, Databricks, Kafka, Kubernetes. 6 months. $10K/month AWS bill. Constant debugging.
Path 2: Master fundamentals. Clean data + solid SQL. Week 1. $50/month. It just works.

😲 What's the difference?
Path 1 looks impressive on LinkedIn. 
Path 2 actually solves problems.

Complexity without fundamentals = expensive tech debt. 
Simplicity with strong basics = reliable systems.

𝘚𝘰𝘮𝘦𝘵𝘪𝘮𝘦𝘴 𝘵𝘩𝘦 𝘣𝘦𝘴𝘵 𝘥𝘢𝘵𝘢 𝘱𝘭𝘢𝘵𝘧𝘰𝘳𝘮 𝘪𝘴 𝘵𝘩𝘦 𝘰𝘯𝘦 𝘺𝘰𝘶 𝘥𝘰𝘯'𝘵 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘣𝘶𝘪𝘭𝘥.

You wouldn't build a skyscraper on quicksand. So why are you building ML models on dirty data?

Data quality isn’t a tooling problem.
It’s a people problem.

You can buy every check, test, and dashboard on the market and still ship bad data if ownership is fuzzy.
Quality only works when everyone treats it as their job.

Here's the reality: Quality is a team sport. Whether you're:
Building pipelines (Data Engineers)
Analyzing trends (Data Analysts)
Training models (Data Scientists)
Enforcing standards (Data Stewards)
Making decisions (Business Leaders)

𝘌𝘷𝘦𝘳𝘺𝘰𝘯𝘦 𝘰𝘸𝘯𝘴 𝘥𝘢𝘵𝘢 𝘲𝘶𝘢𝘭𝘪𝘵𝘺. Not just the data team.

The Six Pillars That Hold Your Data Together:
Accuracy - Does it reflect reality?
Completeness - Are critical fields missing?
Consistency - Same format everywhere?
Timeliness - Fresh or stale?
Validity - Follows defined rules?
Uniqueness - Duplicates eliminated?

❗️𝗕𝗿𝗲𝗮𝗸 𝗼𝗻𝗲 𝗽𝗶𝗹𝗹𝗮𝗿, 𝗯𝗿𝗲𝗮𝗸 𝘁𝗵𝗲 𝗲𝗻𝘁𝗶𝗿𝗲 𝘀𝘆𝘀𝘁𝗲𝗺.

As a data engineer building governance always ensure that you track:
 • Time to Data - How long until new users can access data they need?
 • Data Trust Score - % of users who trust the data they use
 • Incident Response Time - How fast can you fix data quality issues?
 • Self-Service Adoption - Are users finding data themselves or asking you?
 • Compliance Coverage - % of sensitive data properly classified and protected
 • Metadata Completeness - How well documented is your data?

Bad data is expensive in quiet ways:
 Wrong numbers → wrong decisions
 Messy records → weak models
 Schema changes → broken pipelines

Netflix didn’t struggle because of algorithms.
They struggled when late or wrong data changed what people saw.
Data quality isn’t a project you “finish.”

It’s a habit you keep.
The teams that win aren’t drowning in data.
They’re clear on ownership, standards, and trust.

Monitor at every stage—source, pipeline, warehouse—not just at the end.

Why does this matters more than ever?
Data Quality Engineer roles: growing 42% annually
Chief Data Officer positions: $200K-$400K+
GDPR/CCPA compliance is non-negotiable
Gartner predicts 30% higher ROI with formal governance by 2026

Read more on Data Governance roles and responsibilities by Piotr Czarnas - https://lnkd.in/gy9jh_Cr
Refer this to explore what Atlan says about Data Governance pillars in 2026 -https://lnkd.in/dAMTYJGy

𝗚𝗮𝗿𝗯𝗮𝗴𝗲 𝗶𝗻 = 𝗚𝗮𝗿𝗯𝗮𝗴𝗲 𝗼𝘂𝘁. 
But here's what most people miss: Garbage tolerated = Trust destroyed."
What's your biggest data quality challenge? 👇

The GAP is real
Every common person is equally concerns about AI.
"Instagram/ Siri is hearing me and changing the recommendations"

One of my niece said this to me yesterday.
And.. I really felt like we're just on the different sides of the screen.

The irony?
We pull all-nighters worrying about data security...
They pull all-nighters worrying about data security.

What we're actually doing?
 → Building trust & safety frameworks
 → Implementing data privacy by design
 → Creating quality checks on quality checks
 → Making AI explainable and auditable

What people think we're doing?
 → Building Skynet
 → Stealing their job
 → Teaching Alexa to spy on them
 → Making everything too complicated on purpose
 
To get the real time understanding, these are some amazing blogs to read:
1. Seattle Data Guy - https://lnkd.in/g6kC-9wa
2. Data Engineering Things - https://lnkd.in/gg_hCpk7
3. DataExpert.io - https://lnkd.in/gxj-wAHD
4. Confessions of a Data Guy - https://lnkd.in/gXnAFW8g
5. Netflix Tech - https://lnkd.in/gUMZ9xpS
6. Engineering at Meta - https://lnkd.in/gg7k9pWh
7.  Spotify - https://lnkd.in/gJWShJKV
8. LinkedIn Engineering - https://lnkd.in/gw6aBvYn
9. Slack Engineering - https://slack.engineering/
10. Uber Engineering - https://lnkd.in/gyUdJcMb
11. PayPal Tech - https://lnkd.in/gdmiErJY
12. Airbnb Engineering - https://lnkd.in/gQVYUeJY
13. Databricks - https://lnkd.in/gD24FB63
14. Canva Engineering - https://lnkd.in/gYA8eUnQ
15. Jesse Anderson - https://lnkd.in/gYaS58E4
16. Andreas Kretz - https://lnkd.in/gWTNDwUg
17. DataCamp - https://lnkd.in/gEen6W7U
18. Data Engineering Weekly - https://lnkd.in/g7CA7tHM
19. The Analytics Engineering Hub - https://lnkd.in/gyMrwzPM
20. Atlassian Engineering - https://lnkd.in/grvHZskQ

The gap isn't just technical - it's educational, emotional and human.
Let's make AI less scary and more simplified as a support system.

Engineers can build amazing things. 
But we also need to bridge the gap between "𝗵𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀" and "𝘄𝗵𝘆 𝘆𝗼𝘂 𝘀𝗵𝗼𝘂𝗹𝗱 𝗰𝗮𝗿𝗲."

So, where are you right now — building it, learning it, or still figuring out what all the hype is about? 💬

taplio

The Battle

Harpreet Sahota 🥑

Pooja Jain

Followers

Posts

Engagements

Likes

Performance ScoreHarpreet Sahota 🥑's LinkedIn Performance

Performance ScorePooja Jain's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Harpreet Sahota 🥑

Pooja Jain

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);}Harpreet Sahota 🥑's LinkedIn Performance

Performance ScorePooja Jain's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreHarpreet Sahota 🥑's LinkedIn Performance