Vedam School of Technology

• Leading the acquisition, curriculum, strategy, and student experience for India's first AI-enabled college, Vedam.
• Collaborating with stakeholders to develop innovative solutions for student success and growth.
• Implementing cutting-edge technology to enhance the learning experience and drive student engagement.

Assistant Vice President

Tayyari

Building a AI-Tech platform to solve problems faced by students while preparing for competitive exams.

Co-Founder

YouTube

Digital Creator

AlgoPrep

( Acquired by Sunstone )

• Co-founded AlgoPrep, a platform bridging the gap between college education and industry requirements through courses and community.
• Taught over 5000 students and built a committee of more than 100k students to enhance industry readiness.

Microsoft

Working on Live Persona Editor (LPE) and Live Persona Card (LPC) for Microsoft Teams and Outlook.

Software Engineer

Worked on Microsoft Autofill Team to provide world-class user experience in autofill domain (passwords and addresses). Responsible for enabling the most customer asked features like Add Passwords Manually and Improved the overall Import Experience in Microsoft Authenticator iOS App.

Crosslinks NSUT

President

Executive committee member at Crosslinks, NSUT - The public and student relations society of NSUT.

Executive Committee Member

CodeChef

Educator

A.R.E.S. Robotics

Senior Software and Autonomous Traversal Engineer 

Management Lead at A.R.E.S. Robotics. I help in keeping coordination between all the departments and take care of the funds for building the rover.

Management Lead

The Alliance - NSIT's Newspaper

Handling all the day to day activities of the society. Managing the workflow and a team of 40 members.

Managing Editor

Senior Caricaturist

Caricaturist 

Samsung Electronics

Worked in the Network Modem R&D team of SRIB.

Software Engineer Intern

Building Tayyari! 🚀

I am an Ex Software Engineer @Microsoft, Graduated from NSIT. 

I have interned at Samsung Research Institute Bangalore in the Network R&D team and I have won Smart India Hackathon'20. 

I like to explore new things and in this journey i have created a YouTube community with over 400k Subscribers.

I left Microsoft  2 years back.

Getting an official invite to the Microsoft AI Tour in Bengaluru felt amazing 🔥

Witnessed Satya Nadella announce $17.5B investment in India by 2029 and the goal to upskill 20 Million Indians in AI.

But here's what hit me the hardest:

Tata Consultancy Services , Infosys , Wipro , Cognizant each deploying 50,000+ Microsoft Copilot licenses. That's 200,000 licenses across just these 4 companies.

Your company is betting big on AI. Are you?

These aren't just numbers in a PPT. These are seats being filled by people who decided to learn AI tools while others were debating if "AI will take jobs."

The updates coming to GitHub and Microsoft Copilot  are insane. The developers who master these tools will literally 10x their output.

India is getting the investment. The jobs are being created. The infrastructure is being built.

Question is will you be ready when the opportunity knocks?

Excited for what's coming. Thanks Microsoft for the invite ❤️

Also met a lot of exciting folls! 

#microsoftaitour #microsoft #ai #developer

Nishant Chahar

EF (Fall’25) | Building in Stealth | Ex-Algoprep (Acquired) | Ex-Microsoft | 550k+ Subs on YT

No alternative text description for this image

Landed in Delhi, cab to college, 10 minutes until I speak to 500 students. Zero slides ready.

Was invited for a guest lecture about AI tools for students. Flight got delayed, landed with barely 30 minutes to reach campus.

Sitting in the cab, realized I had nothing to present. Just my notes in a Google Doc.

The Panic: 
• Lecture starts in 10 minutes 
• Slides prepared: Zero 
• Audience: 500 students expecting a professional session

What I Did:
Opened Gamma on my laptop, pasted my talk outline, and hit generate.

How to create a presentation in a moving cab:
1. Open gamma(.)app on your phone/laptop
2. Click "Create new" → "Generate with AI"
3. Paste your topic or outline
4. Let Gamma generate slides
5. Quick review and customize if needed

By the time the cab reached the college gate, I had a complete 15-slide presentation covering: 
• 5 AI tools students need to know 
• Practical examples for each tool 
• Real student success stories 
• Action steps they could take immediately

Walked into the auditorium, connected my laptop, and presented like I'd spent hours preparing.

Students asked for the slides afterward. Multiple professors wanted to know how I created such polished visuals so quickly.

Nobody knew I made it 5 minutes before walking on stage.

This Works For: 
• Last-minute client pitches 
• Unexpected speaking opportunities 
• Emergency team presentations 
• Interview case study presentations

The difference between looking unprepared and looking professional isn't hours of work anymore. It's knowing the right tools.

Try Gamma: https://lnkd.in/dEi3Stq7

💾 Save this - For the next time you're in a last-minute situation 
🔄 Repost - Help someone who panics before presentations

#PublicSpeaking #AITools #StudentLife #Productivity #madewithgamma #gammapartner

It's 7 AM, your hackathon is at 9 AM, and you're still fighting with PowerPoint slide templates.

You spent 8 hours building your solution. Now you're wasting 3 hours making slides look "professional."

Meanwhile, your demo is half-baked and you haven't practiced your pitch once.
Hackathon judges give you 5 minutes. They want to see your solution and hear your thinking.

Nobody's giving extra points for slide animations or perfect font pairing.
But students keep doing this:
 • 70% time on building
 • 25% time on slides
 • 5% time on pitch practice

Then wonder why they lost to a team with a weaker product but a stronger demo.

What Winning Teams Do:

For Hackathons:
 • Build your solution → Create pitch deck in 10 minutes with Gamma
 • Spend 3 hours practicing demo and answering questions
 • Walk in confident while others are still exporting PPTs

For Class Projects:
 • Upload your project report → Get presentation slides
 • Focus on understanding your work, not designing slides

For Competitions:
 • Case competitions → Professional deck in minutes
 • Tech presentations → Visual showcase without the hassle

How It Works:
1. Go to gamma(.)app
2. Either generate from topic or upload your document
3. Get professional slides instantly
4. Actually practice your pitch

Time Saved = Better Results

Try Gamma: https://lnkd.in/gH9QWxRQ

💾 Save this - Next hackathon is closer than you think 
🔄 Share with your team - Win together

#Hackathons #StudentLife #CollegeTips #madewithgamma #gammapartner

Entrepreneurs First environments hit different: Someone overheard us, made one intro, we got a healthcare pilot 

After landing our first edtech pilot (EP14), we needed a healthcare customer.
We were actively looking. Talking about it constantly.

Then luck happened.

Deepak and I were discussing how to find a design partner in healthcare. Just talking through our options out loud.

Shashank Dixit was sitting right in front of us. He'd interned at a healthtech company. He overheard us and said: "I know someone. Let me connect you."

Within a day, we were on a call.

We explained our idea, converting unstructured healthcare data into hybrid retrieval systems using GraphRAG.

Patient records. Medical reports. Research papers. All connected through relationships, not just similarity.

He was interested. Really interested.

But here's the thing, getting into big enterprises is tough initially.

Especially healthcare. Long sales cycles. Compliance. Approvals.

Then he mentioned a burning problem they were facing right now.
The real problem:

They had invoices to GRN (Goods Receipt Note) matching, but their invoice extraction wasn't working properly.

Unstructured invoice data → Structured database entries.

It was failing. Costing them time and money.

Our decision was to solve this first. Get into the system. Then show them what GraphRAG can do for their bigger data problems.

We shipped an MVP in 2 days.

The POC (Point of Contact) loved it. He even helped us navigate internal approvals.

What we thought would be a quick proof-of-concept turned into over a month of conversations.

Back and forth. Testing. Feedback. More iterations.

Progress was happening, but it was slow.

After a month, we had to make a decision.

Do we keep waiting? Do we pivot our focus? Do we kill this pilot?

More on that hard call in EP16 👀

#IDKSeries #EP15 #healthcare #startupJourney

Your ICP isn't "everyone with the problem." It's "everyone who will actually pay you to solve it" 👀  After naming our product unifydb.ai (EP12), we needed to answer the most important question: 

Who is our Ideal Customer Profile?

We started talking to two very different types of companies.

Customer Type 1: AI/Agent startups

We reached out to companies like Lucio, Juriosphere, August AI, Metadome, and others building AI-powered products.

Question: How are you handling knowledge retrival right now?
Every company had built something. Basic RAG. Some even had sophisticated setups. Some were using NotebookLM. 

Their accuracy? Decent. Good enough for their current needs.

Knowledge graphs give you an accuracy bump. But it comes with cost and complexity.

Their response? Not interested. They'd already solved it "good enough."

We got a reality check, They needed better RAG, but not badly enough to pay for it.

Customer Type 2:  Traditional enterprises (Edtech, Healthcare, Legal, Finance, Insurance)

We reached out to old-age companies where tech is secondary to the main business.

What we found:
They knew about RAG. Most weren't using it. Two reasons:
  1. Not a priority
  2. Didn't have the technical depth to build it

We decided to focus only on reason #2.

Companies that wanted RAG but couldn't build it themselves—this was our ICP.

The validation:
During our external adviser check-in with Mayank, he supported this thesis.
Traditional enterprises with:
 - Rich, unstructured data (legal docs, patient records, course content)
 - Real need for AI/RAG
 - Limited technical teams
That's who we'd build for.

ICP Defined:
Companies where: 
✅ Tech is not their core competency ✅ They have valuable data locked in documents ✅ They want AI capabilities but can't build in-house ✅ They're willing to pay for a solution

Next step: Get real data from potential customers and build a tailored demo.

More on that in EP14 👀

PS: me and Tharun Naik in HSR figuring out life 👀😂

#IDKSeries #EP13

Finally, an idea that didn't die. Now we needed a name and customers 💡

Naming a startup is like naming your child. Except 1000 people already named their kid the same thing 😂

After EP11, we had conviction. RAG infrastructure was the move.

Now came the fun part: naming it.

We didn't want to spend too much time on this. Just pick something and move on, right?
Wrong.

Naming your startup is like naming your child. You overthink every option. You imagine how it'll sound when people say it. You check if it means something weird in another language.

We spent 3-4 hours brainstorming names.

Every single name we liked? Either:

 - A company already existed with that name
 - The domain was taken
 - Both

Frustrating doesn't even begin to describe it.

Finally, we landed on: https://unifydb.ai/

 - Domain available ✅ 
 - Clean, tech-forward ✅ 
 - Actually describes what we do (unifying data) ✅
Done. Moving on.

Next step decide on a basic architecture and build a working prototype.
We gave ourselves 2 days to ship a basic version.

Not perfect. Not polished. Just functional enough to show customers and get real feedback.

While planning the build, we identified our ideal use cases:
Companies where tech is secondary to the main business:
 - Edtech: Course content, student data, curriculum docs
 - Healthtech: Patient records, medical research, compliance docs
 - Legaltech: Case files, contracts, regulatory documents
 - Fintech: Financial reports, compliance docs, transaction data
These companies had massive amounts of unstructured data but lacked the engineering teams to build RAG infrastructure in-house.

Perfect fit.
We started reaching out.

While building, we scheduled customer calls for the upcoming days.
Show them a working prototype and understand their specific pain points in real-time.

No more hypothetical questions. Just: "Here's what we built. Does this solve your problem?"

We had a few calls lined up. Time to validate with a real product, not just an idea.

That's EP12. In EP13, I'll share what happened when we showed customers the first version 👀

PS: I know it's been some time since the last episode; now we'll have one every alternate day!

#IDKSeries #EP12 #startup #graphrag #rag #knowledgeGraph

Your network is your net worth. We got our first pilot customer in 24 hours 🚀

We started reaching out after defining our ICP: traditional companies needing knowledge graphs but lacking technical depth. EP13

First stop: Edtech
I texted three people Ankush , Anshuman and Piyush. 

All three responded immediately. Within hours, we had calls scheduled.

I explained the concept of GraphRAG infrastructure that helps companies make sense of their unstructured data by understanding relationships, not just similarity.

Ankush sir gave us a usecase to pilot, AI Teaching Assistant powered by GraphRAG.

While Traditional RAG retrieves similar content, learning is about connections, not just about similarity.

For Example a student asks: 
"How do I solve 'Find the longest increasing path in a matrix'?"
[9, 9, 4]
[6, 6, 8]  → Answer: 4 (path: 1→2→6→9)
[2, 1, 1]

What GraphRAG understands:
This problem needs three concepts working together:
 - Arrays (2D matrix navigation)
 - Graphs (DFS to explore paths)
 - Dynamic Programming (memoization to avoid recalculation)

And their relationships:
 - Can't do DFS without understanding arrays first
 - Can't optimize without DP

Need all three to solve efficiently
 - Traditional RAG retrieves similar problems.
 - GraphRAG knows: "Student needs Arrays → then Graphs → then DP+DFS combined" and suggests the learning path.

GraphRAG manages to capture these relationships between the courses, problems, and concepts that vector embeddings alone cannot.

This was exactly what Coding Ninjas needed.

When I talked to Piyush he also showed interest and asked me to share the demo once ready. Potential implementation at Sunstone.

Two edtech pilots lined up, real data coming in, real use case validated.

This was a win.

As edtech was moving fast, we also reached out to healthcare companies.

One conversation with a healthcare giant changed everything.
More about that in EP15 👀

PS:  My first live interaction with Ankush sir was approximately 5 years back, I have met him multiple times after that but never got a picture. Next time :D 

 #IDKSeries #EP14

Ideas 1, 2, 3: Dead. Idea 4: Still alive after 12 customer calls 👀

After spending Diwali on customer calls (EP10), something different happened this time.

The idea didn't die.

We did 10-12 customer calls over the Diwali weekend. Each conversation made our conviction stronger.

The pain was real. Companies needed this. And most couldn't build it themselves.

We also had Akshat in our EF cohort. He'd tried building GraphRAG as a service, even ran a pilot with customers.

But he couldn't scale it.
We talked to him to understand why. What broke? Where did it fail? His insights were gold.
After all the calls, we decided: we're building this.

Why this felt different:
All our previous learnings kicked in:
 - The Mom Test taught us to ask better questions
 - Shorya, Vaishnavi, and Sajal helped us identify patterns we'd have missed
 - Deepak's MetLife experience was our unfair advantage

At MetLife, Deepak had worked on a similar problem, building RAG systems for internal documents—and improved accuracy significantly.

He understood the complexity, the trade-offs, the architecture.
This wasn't theory. This was proven experience.

But wait, what even is RAG?
For those who don't know: RAG = Retrieval-Augmented Generation.
It's how you make AI smarter by giving it access to your specific data in real-time.

Here's how it works:
RAG explained simply:
Computers don't understand words. They understand numbers (0s and 1s).
So we convert text into vector embeddings—numbers that represent meaning in multidimensional space.

Think of it like this: "Walk" and "Run" have similar meanings, so their vector representations should be close to each other in space.
You can create embeddings for words, sentences, paragraphs, or entire documents.

When a user asks a question, the system:
 1. Converts the query into a vector
 2. Finds the closest matching vectors in the database
 3. Retrieves that content as context
 4. Feeds it to the LLM for an answer
That's RAG in a nutshell.

What about GraphRAG?
RAG works great for similarity-based retrieval. But what if your data has relationships?

Questions like:
 - "What products depend on this supplier?"
 - "Show me all documents connected to this project"

Vector embeddings can't capture these connections.
That's where GraphRAG comes in.

Graph Databases have existed for decades, but using them for LLM context retrieval is recent—Microsoft Research pioneered GraphRAG in 2024.
Knowledge Graph stores data as nodes (entities) and edges (relationships), making it powerful for connected information.

It's more expensive to build and maintain, but delivers much better accuracy when relationships matter.

Our next steps:
We started deep-diving into both RAG and GraphRAG architectures.
Then we began competitor analysis: Who's building in this space? What are they solving? Where are the gaps?

More on what we found in EP12 👀

That's EP11. Next one drops soon 🚀

#IDKSeries #EP11

Had a startup idea at 11 PM. Website live by 11:30 PM.

You have an amazing idea. You want to test it, validate it, see if people actually care.

But first, you need a website. And that means: 
 - Finding a designer (2 weeks) 
 - Or learning Webflow/Figma (2 months) 
 - Or settling for a terrible-looking page (2 hours, looks it)
So your idea sits in your notes app. For weeks. Months. Maybe forever.

You can build a landing page that looks professional enough that people take you seriously. Fast enough that you can launch today, not next month.

How to Launch Your Idea in 5 Minutes:
 1. Go to gamma(.)app and click "Create new"
 2. Select "Webpage" as format
 3. Describe your idea: "SaaS tool for college students to track assignments and deadlines"
 4. Click generate
 5. Get complete website with sections, layout, and copy

What Gamma Creates:
 - Hero section explaining your idea 
 - Key features with visuals 
 - How it works section 
 - Sign-up form or waitlist 
 - Mobile-responsive design 
 - Professional layout and colors

Stop waiting for the "perfect" website. Launch a good-enough version today. Learn what users actually want. Iterate.

Version 1 doesn't need to be perfect. It needs to exist.

Gamma gets you from idea to live website in minutes, not months.

Try Gamma: https://lnkd.in/gH9QWxRQ

💾 Save this - For your next idea 
🔄 Share - Help someone launch instead of overthink

#StartupLife #BuildInPublic #NoCode #ShipFast #madewithgamma #gammapartner

taplio

Nishant Chahar's Linkedin Analytics

Nishant Chahar

Performance ScoreNishant Chahar's LinkedIn Performance

What is Nishant talking about?

Nishant Chahar's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Nishant Chahar's Linkedin Analytics

Nishant Chahar

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);}Nishant Chahar's LinkedIn Performance

What is Nishant talking about?

Nishant Chahar's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Nishant Chahar's Linkedin Analytics

Performance ScoreNishant Chahar's LinkedIn Performance

Nishant Chahar's Best Posts (last 30 days)