Satyam Namdev

Built and maintained the backend of VoiceCoreIQ, an enterprise contact center platform using Django ASGI, Gunicorn/Uvicorn, and Azure Communication Services. Worked across 18+ Django apps covering call automation, agent management, queue orchestration, analytics, and omnichannel communication (SMS, Email, Twitter). Developed an AI Voicebot integrating Azure OpenAI GPT-4o Realtime with ACS media streams over WebSocket, supporting real-time audio streaming, tool calling, configurable personalities, and token/cost tracking with MongoDB.

Engineered the RAG pipeline using Azure Blob Storage, Azure Document Intelligence, hierarchical chunking, llama-text-embed-v2 embeddings, and Pinecone for semantic search and transcript retrieval. Implemented Human Agent Handoff workflows using Redis and Celery to seamlessly transfer calls while preserving conversation context.

Built real-time infrastructure using Django Channels and WebSockets for live transcription, notifications, AI assistant interactions, and metrics streaming. Designed async processing pipelines with Celery, Celery Beat, and Redis TLS for call workflows and background tasks.

Designed a multi-tenant MongoDB architecture with MongoEngine ODM and analytics pipelines powering dashboards for SLA, sentiment, CSAT, and agent productivity, including PDF export support via ReportLab. Integrated Azure OpenAI GPT-4o-mini for sentiment analysis and summarization, along with Microsoft Graph APIs for Teams notifications and scheduling. Reduced TTS costs by 85% and call setup latency by 70% using Redis and Blob-based caching strategies.

Containerized services with Docker and nginx reverse proxy support for WebSocket/SSE passthrough. Also collaborated directly with clients to scope features, iterate on feedback, and deliver production-ready solutions.

Spearheaded development of an end-to-end async Legal-AI system using FastAPI, Next.js, Supabase, Weaviate, AWS-S3, serving 200+ legal professionals with 99.9% uptime.

Designed Supabase (PostgreSQL) schemas for multi-tenant chat sessions, document collections, and RAG workflows; opti mized to handle 10,000+ session records with sub-100ms query response times.

Delivered LLM function-calling, streaming responses, multi-LLM orchestration, and real-time WebSocket updates for 100+ concurrent users; integrated live web search via Tavily and Exa.

Developed LangGraph-based Entity Extraction Agent with scheme validation and multi-agent orchestration, achieving 92% accuracy in legal entity extraction across 5,000+ documents.

Refactored a tightly coupled Gradio monolith into a modular FastAPI backend,reducing deployment time by 60% and enabling scalable API integration for 3 major law firms.