Imran Ansari

I’m a Senior Full-Stack Engineer with 5 years shipping production platforms. My recent focus is generative AI applications—LLM tool-calling/function-calling, schema-grounded (RAG-style) prompting, and multi-turn chat sessions that are hardened for real-world use.

At Lepton Software (SDE-2), I built a LeptonLLM-powered agentic geospatial assistant with tool-calling for map rendering and dataset selection. I combined schema-grounded RAG-style prompting using per-dataset _info.json with in-browser Pyodide for sandboxed analysis on user uploads, then took features through production hardening across FastAPI + React 19.

I led the frontend rewrite of SmartMarket.ai into a Rio.js extension monorepo, shipping Datasets, Connections, and an Agentic AI Assistant. I also delivered reliability improvements like skew-protection middleware, a headless print pipeline for PDF reports, and a real-API Playwright trust suite to catch backend connection rename durability defects—along with pioneered Playwright visual regression testing.

Previously at Innostax Software Labs Pvt. Ltd., I served as Project Lead, integrating 15+ ShipStation modules and improving API performance and iteration speed while leading a team of 7. I enjoy building end-to-end systems, partnering closely with teams, and turning prototypes into dependable products—whether through production delivery or open-source tools like my Roads Management Insights route registration work.