Aadarsh Pandit

BTech AI Student @ Kathmandu University | Machine Learning & NLP | Research & Projects

[Aadarsh Pandit @AadarshP77]

I spent 30 days analyzing Startup.

It's not about automation. It's about augmentation.

Relying 100% on Startup makes you generic.

But using Startup to speed up drafts by 80%?

You win.

How do you integrate it?

[Aadarsh Pandit @AadarshP77]

Everyone says Llm is the future.

After 3 apps, I disagree.

The bottleneck isn't the API. It's your indices.

We switched to a vector-first approach and query times dropped by 400ms.

Bookmark this. 🔖

[Aadarsh Pandit @AadarshP77]

Sick of AI theory without real projects?

I built a free Microsoft Learn Plan that bridges the gap.

It covers AI basics → NLP → AI Agents using Azure.

Learn it.
Build it.
Ship it.

Start your journey here: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

I spent 30 days analyzing Training a 4B model.

It's not about automation. It's about augmentation.

Relying 100% on Training a 4B model makes you generic.

But using Training a 4B model to speed up drafts by 80%?

You win.

How do you integrate it?

[Aadarsh Pandit @AadarshP77]

Day 58/100 — Reflection

Week 9 done.

Biggest win: Model accuracy hit 95%
Biggest challenge: Data cleaning took forever
Key lesson: Start small, iterate fast

Onward 🚀

#100DaysOfAI

[Aadarsh Pandit @AadarshP77]

Theory is good, but building is better.

I designed a free Microsoft Learn Plan for developers who want to ship.

We cover AI fundamentals, NLP, and deploy AI Agents on Azure.

Connect the dots. Start here: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

Optimize Fed Raises Rates for in production:

1. Use Fed Raises Rates for for the heavy lifting.
2. Cache aggressively. Not caching? You're burning money.
3. Keep state simple.

What's your go-to Fed Raises Rates for stack? 👇

[Aadarsh Pandit @AadarshP77]

After 3 apps, I disagree.

The bottleneck isn't the API. It's your indices.

We switched to a vector-first approach and query times dropped by 400ms.

Bookmark this. 🔖

[Aadarsh Pandit @AadarshP77]

Most AI learning paths leave you stuck between theory and actually building. I wanted one that connects the dots. So I created a free Microsoft Learn Plan covering AI → NLP → AI Agents with Azure. Learn it. Build it. Ship it. Start here: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

https://t.co/WXwwny6azF🔗 Check out the code and all my open-source projects on my GitHub!

[Aadarsh Pandit @AadarshP77]

Day 53/100

A lot of people ask how I broke into AI research.

Build in public.

I shipped 'Smart NID Helper' to solve a real local problem.

I documented the whole process.

Don't just research. Deploy.

[Aadarsh Pandit @AadarshP77]

Hot take:

Claude are more important than the models themselves

The best devs are the ones who learn to use AI as a force multiplier

Fight me on this. 👇

#SoftwareDev

[Aadarsh Pandit @AadarshP77]

Stuck in tutorial hell?

I created a free Microsoft Learn Plan that actually connects the dots.

From AI and NLP straight to building AI Agents with Azure.

Learn it. Build it. Ship it.

Get started for free: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

Rank these (hardest first):

1. System Design
2. DSA
3. ML/AI
4. Frontend

My ranking: 1→3→4→2

What's yours? 👇

[Aadarsh Pandit @AadarshP77]

I spent 30 days analyzing Rag.

It's not about automation. It's about augmentation.

Relying 100% on Rag makes you generic.

But using Rag to speed up drafts by 80%?

You win.

How do you integrate it?

[Aadarsh Pandit @AadarshP77]

Most AI learning paths leave you stuck between theory and actually building. I wanted one that connects the dots. So I created a free Microsoft Learn Plan covering AI → NLP → AI Agents with Azure. Learn it. Build it. Ship it. Start here: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

Everyone says Next.js is the future.

After 3 apps, I disagree.

The bottleneck isn't the API. It's your indices.

We switched to a vector-first approach and query times dropped by 400ms.

Bookmark this. 🔖

[Aadarsh Pandit @AadarshP77]

The hardest part of Llm?

Scaling it.

Here is the fix:

What used to take a whole DevOps team can now be done in 3 CLI commands using Llm.

Stuck? Drop your error below. 👇

[Aadarsh Pandit @AadarshP77]

Sick of AI theory without real projects?

I built a free Microsoft Learn Plan that bridges the gap.

It covers AI basics → NLP → AI Agents using Azure.

Learn it.
Build it.
Ship it.

Start your journey here: https://t.co/47JkK3AlcN #AIAgents

[Aadarsh Pandit @AadarshP77]

The hardest part of Small programming tricks?

Scaling it.

Here is the fix:

What used to take a whole DevOps team can now be done in 3 CLI commands using Small programming tricks.

Stuck? Drop your error below. 👇