UtkarshSingh
I'm a senior software engineer with 11+ years building production systems in financial services — now designing and shipping GenAI and agentic AI features end to end.
Eleven years of backend. Now building agents.
I'm a senior engineer with 11+ years building production backend systems in financial services — trading, money movement, and securities management — including four years delivering event-driven microservices and high-traffic investment platforms for Fidelity.
Now I design and ship GenAI and agentic AI features end to end: tool-calling agents, multi-step orchestration, and RAG with hybrid retrieval. I also lead internal AI enablement — agentic frameworks, the OpenAI SDK, and organization-scale GitHub Copilot adoption.
/ Toolkit
- Languages
- Python, TypeScript, Go, Java, SQL
- GenAI & Agents
- RAG, LangGraph, CrewAI, MCP, LiteLLM
- Backend
- FastAPI, asyncio, Kafka, Microservices
- Infra & Data
- AWS, Kubernetes, PostgreSQL, Qdrant, Neo4j
Recent work in AI and systems.
Code-Analyzer
2025A code-intelligence platform that ingests and queries large multi-repository codebases — 600+ repos at its first engagement — to power software-engineering audits. A seven-layer hybrid pipeline pairs tree-sitter parsing with a Qdrant + Neo4j + OpenSearch retrieval backbone and cross-encoder reranking.
Resume Screener
2025An AI resume-evaluation system on FastAPI that flags exact duplicates, near-duplicates, and semantic plagiarism through a multi-signal pipeline of hashing, lexical overlap, and embeddings — with PII-aware ingestion and human-reviewable evidence reports.
Meeting AI
2024A meeting-intelligence platform that transcribes audio and generates structured minutes, summaries, and conversational Q&A using RAG over timestamped transcript segments, backed by PostgreSQL-native vector search.
Where I've spent my time.
Designing agentic systems that automate enterprise infrastructure and onboarding.
- Built the TerraCore Onboarding Agent — an agentic system using LLM tool-calling and multi-step orchestration to automate environment provisioning and account servicing.
- Exposed the agent as an async FastAPI service with streaming progress, coordinating Terraform, Jenkins, and account-servicing APIs as agent tools.
- Engineered guardrails — input validation, allow-listed actions, dry-run modes, and human-in-the-loop gates — for safe, auditable runs in regulated environments.
Let's build something good.
Have a project, a role, or an idea worth chasing? My inbox is open — I read everything and reply to most.