|
|
name: Krishna Birla role: AI/ML Engineer works_at: ProPlus Data (Jr. ML Associate) based_in: Pune, India building: RAG systems, agent pipelines, backends next: B2B services, Edutech, observing fuel: Coffee > Code > Repeat |
I turn messy data into systems that answer questions. Mostly Python, mostly AI, always with a backend that has to survive real users.
- Agentic AI: multi agent pipelines with LangChain, LangGraph and local models on Ollama
- RAG and fine tuning: retrieval systems on MongoDB Atlas, a fine tuned Llama 3 model at work
- Backend: FastAPI and Flask services that put AI behind a clean API
- Data and scraping: Selenium, Playwright and Apify pipelines that collect data at scale
me explaining to stakeholders why the model needs more data
Jr. ML Associate at ProPlus Data (Jun 2025 to present), working on Nexora.
- Led the backend, AI integration and MongoDB Atlas setup for a custom RAG system
- Fine tuned a Llama 3 model on company data and wired real time web access around it
- Moved the Decision Logic Pipeline from an external API to local models: concurrency went from 1 to 6, processing got 4x faster, and costs dropped
| 4x | 35 | 5 | 47K+ | 8.41 |
|---|---|---|---|---|
| faster processing after going local | agents in my SLM swarm | agents in Founders Intelligence | Q&A pairs in my Medical FAQ RAG | CGPA out of 10 |
|
Founders Intelligence |
Multi Agent SLM Swarm |
|
Verification Engine |
Drone Security Agent |
|
Archiscan AI |
UIDAI Data Hackathon 2026 |
More builds
- Stock Assist: reads balance sheets, cash flows and income statements and explains them in plain English
- Medical FAQ RAG Chatbot: FAISS search over 47K+ medical Q&A pairs, with a Streamlit UI and BLEU and ROUGE checks
- Data Scraping Engine: Selenium pipeline that collects company contact data
- Botnet Detection: Random Forest on the CTU 13 dataset, with a Streamlit dashboard
- Marketing Analysis: sales trends, customer behaviour and product performance with Pandas
actual footage of me when the pipeline finally runs clean
| Languages | |
| AI and agents |        |
| Backend and data |   |
| Infra and testing |     |
- IEEE research paper: Architectural Design by Blueprint Generation using GenAI (May 2026)
- IJSRET research paper: Landslide Prediction Using ML and GIS Based Approach (May 2025)
- Oracle Certified Generative AI Professional (Oct 2025)
- DataKind open source contribution on financial inclusion data for India (Apr 2025)
Education: B.E. in Artificial Intelligence and Machine Learning, ISBM College of Engineering, Pune (2022 to 2026)
The long term plan is autonomous systems with real intelligence. Drones that understand what they see. Robots that use AI as a nervous system, not a feature on a slide. Archiscan AI and the Drone Security Agent are the first steps.
the plan. ambitious? yes. unhinged? maybe.
refreshes on every visit. just like my commitment to fixing that one bug.
find me here
This README was updated after a very long gap. The commit history knows.