Architected and deployed an agentic application using LangGraph and Claude 3.5 Sonnet for dynamic reasoning and multi-step tool invocation; reduced task completion time by 40%. Optimized LLM inference across vLLM, Transformers, and ONNX Runtime to achieve 5x throughput, sub-200ms latency, and 30% cost savings. Built RAG pipelines with LangChain, LlamaIndex, and Pinecone; developed deterministic SQL generation with Jinja templating for enterprise finance data.

Principal Software Engineer | Technical Team Leader at ColomboAI

May 1, 2023 - March 31, 2025

Architected a cloud-native financial analytics platform on AWS (Lambda, API Gateway, DynamoDB) serving 200+ enterprise healthcare clients with 99.95% uptime. Designed event-driven microservices (Node.js/Express, SQS/SNS) processing 5M+ daily healthcare claims with Docker Swarm. Built AI-powered fraud detection (XGBoost, TensorFlow autoencoders) achieving 87% F1 and flagging $15M+ in fraudulent patterns. Optimized REST/GraphQL APIs, caching, and streaming to reduce P99 latency to 800ms. Implemented HIPAA-compliant authentication, CI/CD with Terraform and EKS, and robust observability; reduced deployment cycles to ~8 minutes with zero downtime. Mentored 8 junior engineers, boosting velocity by 35% and reducing production bugs by 42%.

ML /AI Full-Stack Engineer | LLM Developer at Livy AI

October 1, 2020 - April 30, 2023

Led modernization of a legacy SaaS platform into microservices; containerization reduced AWS EC2 costs by 45%. Refactored to 18 event-driven microservices, improving deployment reliability across regions. Built AI-powered document automation (OCR + NLP) processing 2k PDFs/day with 92% extraction accuracy. Delivered a real-time analytics dashboard and improved Core Web Vitals, increasing retention by 22%. Implemented comprehensive test suites; enabled Agile processes and improved team velocity by 35%.

Software Developer at Amazon

July 1, 2016 - September 30, 2020

Contributed to a cloud-native SaaS platform, optimizing REST APIs with PostgreSQL connection pooling and Redis caching to reduce latency under load. Built reusable React UI components and improved Core Web Vitals, enhancing SaaS retention. Supported multi-region deployments and DevOps workflows with Docker and Jenkins; expanded test coverage from 45% to 75% and improved reliability across 10k+ daily active users.