Professional Experience

Backend AI Engineer

ezAIx, IncezAIx, Inc

June 2025 - Present

Thousand Oaks, USA (Remote)

  • •Architected and developed 15+ backend APIs using Django REST Framework (CBVs) to power a customer service call routing platform serving 1,000+ daily calls integrated with Azure Communication Services (ACS).
  • •Deployed periodic background task scheduling using Celery, Celery Beat, and Redis, processing 500+ automated tasks daily with automated backend triggers via Ngrok-exposed endpoints.
  • •Built backend logic to fetch, handle, and route live calls from the ACS grid; implemented queue-based call routing reducing customer wait time by 40% based on real-time customer intents and flow configuration.
  • •Integrated ACS SDK features such as call recording and transcript generation; developed webhooks and WebSocket-based microservices to stream transcripts live to 50+ concurrent users.
  • •Stored and indexed 10,000+ call transcripts in Pinecone to enable semantic search with 95% accuracy and efficient call-transcript mapping, enhancing post-call insights and retrieval for customer service teams.
Django REST FrameworkAzure Communication Services (ACS)PineconeCeleryRedisMongoDBCosmosDBDockerWebSocketsNgrok

Founding Engineer

TunableLabs, LLCTunableLabs, LLC

Nov 2024 - May 2025

San Francisco, USA (Remote)

  • •Led the development of a Legal-AI RAG system with support, using multi-tenant support in Weaviate as the vector database. Added real-time web search features with Tavily and EXA APIs to fetch accurate information. Built AWS S3 integration to handle user document uploads and storage efficiently. Created a pipeline using Unstructured.io to read and break down different types of legal documents into useful sections.
  • •Architected multi-agent workflows using LangGraph (won internal hackathon worth $200), including dynamic entity extraction and schema mapping
  • •Developed secure user access with Supabase Auth integrated into a Next.js frontend
  • •Migrated an AI system from Gradio to FastAPI and NextJs that supports both local and API-based LLMs (OpenAI, Azure, Ollama, Groq), with a hybrid RAG pipeline using vector + full-text search and re-ranking for better results. Added multi-modal QA on documents with text, tables, and images, plus in-browser PDF viewer with citation highlights, relevance scores, and low-confidence warnings. Supported complex questions using breakdown methods and agent-based reasoning (ReAct, ReWOO). Made core retrieval and generation settings.
RAGLangGraphLangChainMulti-agent OrchestrationDockerNext.jsFastAPIUnstructured.ioAWS S3Supabase Auth

Freelance AI Engineer

Self-employed

Oct 2024

Ghaziabad, India

  • •Delivered an ultra-fast SQL-aware chatbot achieving average query response times of < 2 seconds for live data analytics using defog-sqlcoder and Groq-LLM
  • •Enabled contextual schema reasoning and real-time interaction over WebSocket, backed by a secure Supabase data layer
  • •Followed modular architecture with pluggable LLM and SQL generation modules
  • •Ensured maintainability by documenting schema-driven query flows and encapsulating prompt logic
LLMsSQLGroq-LLMAgentic AIWebSocketsSupabaseDefog-SQLCoderModular Architecture