Led technical delivery of 3 production AI software systems, owning architecture, backend services, frontend workflows, cloud deployment, observability, and production support. Engineered a real estate AI assistant with LangChain, LangGraph, Python, React, Pinecone, LLMs, and external data sources, enabling property search, AI chat, lead qualification, visit scheduling, and CRM follow-up management. Orchestrated a 4-agent workflow for property search, user conversation, interest confirmation, and scheduling, guiding users from property discovery to booked visits in one production experience. Launched LLM-powered applications with OpenAI, Claude, Gemini, and OpenRouter, improving response quality, reliability, cost control, and low-latency performance for production users. Increased AI answer relevance by 30% through RAG pipelines with Pinecone, FAISS, Chroma, and Qdrant, grounding responses in structured and unstructured business data. Automated daily activity reporting across 5 sources - Slack, Google Drive, GitHub, chat history, and internal data - using React, Django, Python, SQL, REST APIs, LangChain, and background jobs. Architected scalable backend services with Python, FastAPI, and Node.js for AI-powered SaaS platforms, improving delivery consistency across product workflows. Delivered 4 responsive frontend dashboards and user-facing workflows using React, Next.js, JavaScript, and TypeScript, shortening internal review time by 25%. Strengthened deployment reliability by containerizing services with Docker and managing cloud releases through AWS, Azure, GCP, and CI/CD pipelines. Cut production debugging time by 35% by implementing structured logging, Sentry monitoring, and error tracking for faster issue detection and maintenance. Mentored 2 junior developers during project handover, documenting architecture decisions, support procedures, and production maintenance workflows.
