I work at the intersection of applied AI engineering and backend systems: agentic workflows, RAG pipelines, evaluation, retrieval, APIs, latency, and reliability.
I am looking for roles around AI Engineering, Applied AI Engineering, AI Software Engineering, and Backend Engineering for AI products.
| Signal | Proof |
|---|---|
| Agentic systems | LangGraph support triage, multi-agent supply-chain monitoring, evaluator work in Google ADK |
| Retrieval and grounding | BM25 + dense retrieval, reranking, citation validation, source-linked answers |
| Backend depth | FastAPI/Flask/Node APIs, async processing, PostgreSQL/Supabase, MongoDB, Cloud Run, AWS EC2 |
| Competitive programming | CodeChef #6 India / #76 Global, LeetCode 2665 Guardian |
| Open source | 21 PRs counted across merged and active work in AI, data, devtools, and infrastructure repos |
| Project | Why it matters |
|---|---|
| Grounded Support Triage | Multi-domain support agent with LangGraph routing, hybrid retrieval, deterministic citation checks, escalation fallback, traces, and eval tooling. |
| EdgeSentinel | Edge-native bot detection using Cloudflare Workers, Durable Objects, KV, D1, Vectorize, Workers AI, and OpenAI fallback. |
| Lumina Search | Hybrid RAG/search engine with FAISS, BM25, SQLite content-hash caching, AsyncIO backend, and transparent scoring. |
| SupplySentinel | Autonomous multi-agent supply-chain risk monitor with confidence loops, deduplication, and Cloud Run deployment. |
| TurboQuant KV Cache | PyTorch implementation of KV-cache compression with reproducible benchmark artifacts and quality/throughput tradeoff reports. |
| PinPoint AI | Flutter logistics app with exact pins, saved locations, voice notes, push notifications, agent claiming, and AI-assisted rider guidance. |
More public work: Antim, Lecture Voice-to-Notes, AI Network Traffic Shaper, College Lost & Found, UML Designer AI, productionize.
| Repository | Contribution |
|---|---|
| mem0ai/mem0 #3549 | Azure AI Search vector store support for the TypeScript SDK. |
| pandas-dev/pandas #62592 | PyArrow datetime merge regression coverage. |
| sqlfluff/sqlfluff #7163 | PostgreSQL OPERATOR() syntax support. |
| interviewstreet/hiring-agent #133 | PyMuPDF memory-leak fix using document context management. |
| google/adk-python #3503 | Open PR adding non-English evaluator support. |
| kubeflow/sdk #436 | Open PR improving real-time log streaming behavior. |