Hi, I'm
AI/ML engineer
I build AI systemsthat keep workingwhen things go wrong.
AI Engineer at Easehawk Technologies, and in my final year of B.Tech at VIT. From Bangalore.
[ pick a card ]Bangalore ·
Production AI01 / 04
Easehawk
5
live products I own the AI on
Sole owner of the AI layer, since May 2026.
May 2026 – now
AI Engineer
Easehawk Technologies · Bengaluru
Sole owner of AI architecture, prompt design and production deployment across five live products: Moxsend, MoxReply, MoxCRM, MoxCampus and the EaseHawk avatar chatbot.
[ see the work ↓ ]2024 – 2025
Technical & Research Team Lead
E-Cell, VIT · Vellore
Automated 100+ live data sheets for a flagship event with 5,000+ applicants, with zero downtime across 8 hours. When the tech team couldn't deliver on the day, I built and deployed the event app in under 24 hours.
[ education ]
2023 – 2027
B.Tech, Information Technology
Vellore Institute of Technology · Vellore
GPA 8.56 / 10, kept while working full-time at Easehawk. Graduating June 2027.
[ 01 / 04 ]Easehawk Technologies · since May 2026
A cold email platform up against Lemlist, HubSpot and Salesforce.
I own the AI: a personalised email for every lead, tracking of replies and engagement, and an inbox that reads the replies and flags the leads worth following up.
Python, FastAPI, LangChain, Qdrant
[ owner's afternoon ]
[ 02 / 04 ]VIT, 2026
A receptionist for small businesses, and a guide for their customers.
A chat-first booking app for both sides. Customers get places that fit what they like, with the free slots at each. Owners get a hands-free receptionist that asks them to accept a booking, then asks when they'll be free.
LangGraph, FastAPI, Next.js, MongoDB
[ illustrative simulation · comfort when occupied, setback when empty ]
[ 03 / 04 ]Honeywell hackathon
Agents that run a building's climate for the people inside it.
Most buildings follow a fixed schedule. EcoLoop uses agents and MCP tools to see who is actually in each zone, then adjusts temperature, humidity, filtration and CO₂ for them.
Python, MCP, Groq, EnergyPlus, Streamlit
[ 04 / 04 ]Published research
A model of how one person speaks.
Evoke learns the words someone uses and how they say things, so you can talk to a person you miss. I published a research paper on it.
Schema-conditioned RAG, AWS serverless, Bedrock, LangChain, ElevenLabs
[ What I build with ]
[ Where I'm strongest ]
Building a whole AI pipeline myself, without a framework to lean on. When you write every piece, you know every way it can break.
[ What I'm learning now ]
Production backend and system design. If the data going in and out of an AI pipeline isn't handled well, it doesn't matter how good the model is. It falls over in production.
[ The full list ]
A few I couldn't stop thinking about:
I'll admit it sometimes means I end up solving something that wasn't in the requirements doc. I've made my peace with that.
01 · School years
Football was the first thing I took really seriously. It taught me teamwork, how to deal with hardship, and how to lose and still come back for the next game.
| School teamcaptain for the last 2 | 6 yrs |
|---|---|
| Clubspicked me up along the way | 3 |
| Karnataka U-15All India nationals | 2nd |
| Karnataka U-17All India nationals, as captain | 1st |
02 · Class 10
It started in 10th grade with the calculator program everyone writes in school. Mine was nothing special, but I saw how big computing is, and I understood early why computers were built in the first place. I've wanted to build with them ever since.
[ try it. the keyboard works too ]
03 · College, E-Cell
At E-Cell I handled the technical side of our events. For one of them I'd worked out how it should run, and the team couldn't get it built. I'd made a rough version myself, just to show them what I meant. On the day of the event, that's what we ran.
That's when I stopped thinking of tech as something I was studying, and started building my life around it.
[ event day ]
200
people used it, live
04 · May 2026
I joined Easehawk Technologies with no industry experience. My projects were the closest thing I had. Now I own the AI on their live products.
Say I'm choosing an LLM API. Before I write any integration code I check what it costs, how fast it responds, and how it does on my own test cases. That's slower at the start, but when something breaks later I usually already know where and why.
[ Learned from the bottom up ]
A trigram model I wrote from scratch, trained only on the words on this page. It's small on purpose, so you can watch next-word prediction happen one step at a time.
start with
Honestly, I don't always know how I juggle college, work and everything else. What I do have is a diary. Everything I need to do goes in it with a deadline, whatever kind of day I'm having, and it gets done before the deadline.
I'm proud that I can slowly support myself now. That one means the most to me.
In the next few years I want to get really good at one thing, AI/ML, and be the person trusted to make the big calls with it, the ones that shape a team or a company. I don't want to only be learning. I want to be leading too.
I'd also like to have built two or three products that people keep paying for, because they actually fix their problem.
I'm looking for a full-time AI/ML role. Remote suits me best, but I'm happy to be on-site. If you've got a problem nobody's put into words yet, I'd like to hear about it.