Languages: Python · JavaScript · TypeScript · C++ · SQL
Frontend: React · Next.js · Tailwind CSS · Vite · Redux
Backend: FastAPI · Flask · Node.js · Express.js
Databases: PostgreSQL · MongoDB · MySQL · Redis · SQLite
AI / LLM: LangGraph · LangChain · OpenAI API · Google Gemini · Hugging Face · RAG
ML / Data: PyTorch · TensorFlow · Scikit-Learn · XGBoost · Pandas · NumPy
CS Fundamentals: DSA · OOP · DBMS · Operating Systems
Tools: Git · GitHub · Postman · REST APIs
DevOps/Cloud: Docker · GitHub Actions · AWS · Vercel · Cloud ComputingIdeas are easy. Systems that hold up under pressure are the fun part.
Whether it's fusing multi-instrument spacecraft data to forecast solar flares, building a pipeline that catches an attack in real time, or shipping a full-stack app end to end — I'm happiest once something stops being a plan and starts being a thing that actually runs. This is roughly what that looks like, every time:
$ ./build.sh --project="whatever I am working on this week"
[1/4] PROTOTYPE ........... done # get something running, fast, ugly is fine
[2/4] BREAK IT ............ done # find the input it cannot handle
[3/4] HARDEN .............. done # fix it, then go looking for the next one
[4/4] SHIP ................ done # only counts if it survives contact with reality
$ cat focus.json
{
machine_learning: "training models, breaking them, fixing them",
applied_science: "grounding ML on physics, not just curve-fitting",
software_engineering: "code I will not be embarrassed to reread in 6mo",
full_stack: "backend logic + frontend polish, both matter",
open_source: "reading other peoples code, surviving merges",
problem_solving: "DSA grind, fundamentals do not expire"
}
$ echo "status: still running"🌱 Open Source — GSSoC contributions: PR-ready fixes for race conditions, auth-before-accept vulnerabilities, and distributed-locking bugs, plus senior-level architectural audits across half a dozen codebases that aren't grading me.

