I build agentic systems with a real backend: tools, memory, checkpoints, and eval — not chat wrappers. Also inference measurement (ONNX, quantization) when the loop needs a smaller model.
Resume · Portfolio · LinkedIn · Microsoft Learn · LeetCode · Codeforces · Kaggle · Email
This year: 45 contributions · 20 commits · 13 PRs · 1 review.
Achievements:Pull Shark · YOLO · Quickdraw
Upstream repos I contribute to:
- microsoft/onnxruntime
- langchain-ai/langgraph
- microsoft/autogen
- microsoft/agent-framework
- future-agi/future-agi
Agent runtime + serving path I actually use:
| Area | Tools |
|---|---|
| Agents | LangGraph, AutoGen, tool calling, checkpoints, RAG + FAISS |
| Backend | FastAPI, PostgreSQL, Redis, Docker, GitHub Actions |
| Inference | ONNX Runtime, vLLM, llama.cpp, GPTQ / AWQ (when measuring, not as the product) |
| LangGraph | Pregel / DeltaChannel persistence, checkpoint rebuild, HITL |
| AutoGen | Multi-agent plan → tools → memory |
| InferLite | Runtime + quantization bench (TTFT, tok/s, memory) |
| ONNX Runtime | WebGPU 2-bit quantization path |
LeetCode1708 · 525 solved · CodeforcesShivani_103
B.E. ECE, UIET, Panjab University (2021–2025) · CGPA 8.09/10
Microsoft Learn · Responsible Generative AI · GSSoC