RAG-Based
Document Q&A
Sub-second semantic search across multi-document knowledge bases, with every answer traced back to a source chunk.
/ 00 machine learning engineer gurugram, in
I’m Rohan Chaudhary — a Computer Science undergrad and ML engineer building grounded RAG systems, predictive models, and fast FastAPI backends that hold up outside the notebook.
01 Top 25% on LeetCode
02 4x ML / AI certified
03 Open to internships & roles
/ 04 tiny lab / browser game
A tiny attention test for humans. Find the live vector before it moves. No tracking, no leaderboard — just a 20-second inference loop.
Click the neon target as it jumps through the embedding grid. Every hit increases your streak; every miss costs one.
Press start to initialize the grid.
Use the arrow keys or the pad to collect data chunks. Grow the sequence, avoid the walls, and see how long your pipeline survives.
Press start, then route the signal.
Adjust gravity, launch the probe, and watch a live vector body bend around the central mass. A tiny two-body experiment for your browser.
Set gravity, then launch the vector.
/ 01 selected systems
Three applied systems, built around a simple question: can the model make a decision more useful, more explainable, or faster?
Sub-second semantic search across multi-document knowledge bases, with every answer traced back to a source chunk.
A Zomato / Zepto-style ETA engine factoring distance, traffic, weather, and kitchen prep time.
Upload a CSV, get automated cleaning, parallel classifier training, and interpretable model charts.
/ 02 engineering stack
The strongest systems are rarely one-model stories. I work across data, modeling, APIs, and the glue that gets a system into the hands of a user.
/ 03 field notes
Why chunking, reranking, and source traceability matter more than another prompt tweak.
6 min read ↗ 02APPLIED MLFrom Haversine distance to prep-time drift: feature engineering for last-mile prediction.
8 min read ↗ 03CAREER / CODENotes from shipping APIs, solving contests, and making model behavior legible to the next engineer.
5 min read ↗/ 04 open source / competitive code
Beyond shipped projects, I keep learning in public: small utilities, algorithmic solutions, and experiments that make the next system a little sharper.
Visit GitHub ↗/ 05 context & credentials
B.Tech — Computer Science & Engineering
Gurugram, Haryana · CGPA 7.0 / 10.0Higher Secondary Education, C.B.S.E Board
Dehradun, U.K. · Score 66%10th Matriculation, RBSE Board
Bharatpur, Rajasthan · Score 96%/ 04 competitive programming / problem solving
I turn messy problems into clean invariants, then ship the solution. Follow the practice trail across LeetCode and Codeforces.
Focused practice in arrays, graphs, dynamic programming, binary search, greedy reasoning, and complexity-aware implementation.
/ 06 start a conversation
Tell me what you’re building, where the bottleneck is, or what you’re curious about. I’m currently open to ML / software engineering internships and roles.