ML & Analytics
PlayerPulse — Churn Prediction & Retention
Leakage-safe churn prediction and retention analytics for mobile games, built on the real Cookie Cats A/B-test dataset (90k players).
Mostly here for the lore (and love) of making machines learn — and the drama that comes with it. If you'd rather meet me through a website than a resume, keep scrolling: everything below is stuff I've actually built, shipped live, and debugged well past the 22nd try.
ML & Analytics
Leakage-safe churn prediction and retention analytics for mobile games, built on the real Cookie Cats A/B-test dataset (90k players).
LLM Infrastructure
OpenAI-compatible API gateway with distributed rate limits, budget caps, automatic provider fallback, and full Grafana observability.
ML & Analytics
XGBoost podium probabilities for every driver in any 2021–2024 Grand Prix, with live what-if scenarios, real weather, and SHAP explainability.
GenAI + Data
Natural-language SQL interface with AST guardrails, hallucination detection, and evidence-based confidence scoring on every answer.
ML & Quant Finance
Stock forecasting, RL trading agents, and Monte Carlo simulations on a modern PyTorch stack — with the statistical rigor most repos skip.
Geospatial ML
Phenology-aware moisture-stress index and FAO-56 irrigation advisory from Sentinel-1/2 satellite data, for canal command areas in Punjab.
Computer Vision
Real-time webcam eye-strain monitoring: a 96.57%-accurate CNN fused with geometric blink detection, live strain index, and a React dashboard.
Full-Stack
Collaborative mini Trello/Linear — MERN + Socket.io with live multi-user sync, RBAC, and optimistic-concurrency conflict resolution.
Full-Stack
Real-time HLS stream monitoring with live VU meters, health scoring, FFprobe analysis, and alerting — built at Amagi Media Labs.
GenAI + Fintech
15-layer transaction data-quality engine blending deterministic rules, Isolation Forest anomaly detection, and guardrailed Gemini summaries.
I learn best with a terminal open and something half-broken in front of me — the gap between “it works on my machine” and “it works” is where I live.
Currently in my third year of AI & ML at BMS College of Engineering, Bengaluru. Lectures hand you the map, but the actual territory — the messy datasets, the models that quietly lie, the deploys that fail at midnight — only shows up when you build things and ship them. So that's my routine: pick an idea slightly out of my depth, wrestle it into working, and walk away knowing more than any textbook chapter could have taught me.
Away from the keyboard, I'm usually a few chapters deep in a good novel — recovering from whatever bug survived my first twenty-one attempts.
On the hunt for AI/ML internships and teams that ship things for real users. Whether you've got an opportunity, a messy dataset, or just a hard problem worth arguing about — my inbox is open.