I design and ship microservices-based AI systems end to end — LLM orchestration, RAG, embeddings, async processing, and containerized deployment on cloud infra. Most of it runs in production behind CI/CD, handling real traffic with real reliability requirements.

• 🔎 RAG & semantic retrieval — vector search over document knowledge bases (Chroma, sentence-transformers, LangChain), tuned for accuracy and latency.
• 🤖 Multi-agent systems — intent routing to specialized agents, session + long-term context across Redis and MongoDB.
• 📄 Document intelligence — Excel/PDF extraction with OCR, field-level confidence scoring, and queue-driven async processing.
• 🎙️ Real-time voice agents — low-latency assistants over WebRTC using the OpenAI Realtime API + Whisper, wired to real tools.
• 🧩 Matching engines — cascaded TF-IDF / cosine / fuzzy matching linking messy free text to structured catalogues at scale.
• 🛡️ Resilient services — AI circuit breakers, retries with backoff, SSE streaming, multi-env Docker + GitLab CI/CD.

Languages

AI / ML & LLM

Backend & APIs

Data & Storage

Cloud & DevOps