Faaraan Kazi

About Faaraan Kazi

I\'m an AI Engineer who specializes in building autonomous agents and agentic systems that are production-ready, not just proof-of-concept.My work sits at the intersection of LLM orchestration, cloud infrastructure, and software engineering designing systems where multiple AI models collaborate, reason through complex tasks, and operate reliably at scale. I work extensively with LangGraph and LangChain for stateful agent workflows, MCP Server for tool integration, and multi-provider model orchestration across Azure OpenAI, Anthropic Claude, and Google Gemini.Where I go deeper than most- Agentic architecture: Multi-model tiering, dynamic tool selection, Human-in-the-Loop (HITL) interrupt design, and context management at the middleware level - RAG systems: hybrid vector/keyword retrieval with Azure AI Search, graph-based RAG with Cosmos DB Gremlin, and retrieval pipeline optimization for accuracy and latency - Azure AI infrastructure: Azure Container Apps, Azure AI Foundry, Azure Service Bus, PostgreSQL state persistence, and Bicep IaC for end-to-end deployments - Observability: Structured logging, OpenTelemetry tracing, LangSmith integration, and token-level usage tracking in production agent systems - Full-stack delivery: Python FastAPI backends with async streaming, Next.js/React frontends with real-time SSE, from architecture to deployment I also bring domain experience in banking, finance, and enterprise automation including fraud detection models, analytics dashboards, and data governance frameworks which means I understand the operational constraints and compliance context that enterprise AI systems must work within. I\'m most valuable on teams tackling the hard parts of AI engineering: reliability, cost efficiency, human-AI collaboration, and turning frontier model capabilities into systems that actually run in production.

Experience

1. Ai Engineer

   Capgemini

   Apr 2025 — Present · Atlanta, GA, US

   Design and deploy production-grade autonomous AI systems and agentic workflows for banking and enterprise clients, with a focus on reliability, scalability, and measurable business impact. Key Contributions- Autonomous Agent Development: Architected multi-agent LLM systems using LangGraph state machines and LangChain orchestration, with Human-in-the-Loop (HITL) interrupt design, dynamic tool selection, and per-turn model tiering across frontier, standard, and economy models (GPT-5.1, Claude, Gemini)- Azure AI Infrastructure: Designed and deployed AI applications on Azure using Container Apps, Azure AI Foundry, Azure Service Bus, and PostgreSQL for stateful agent checkpointing — fully provisioned via Bicep IaC - RAG Pipeline Engineering: Built hybrid retrieval pipelines combining Azure AI Search (vector + keyword) and Cosmos DB Gremlin graph traversal, improving retrieval accuracy by 30% and reducing response time by 40%- MCP Server Integration: Developed MCP-based tool integration layer enabling domain-specific code generation for Temenos banking specifications, reducing development effort by 35%- Multi-Provider LLM Orchestration: Implemented a model registry supporting Azure OpenAI, Anthropic Claude via Azure AI Foundry, and Google Gemini — with dynamic routing, cost optimization, and reasoning effort control per task phase - Observability & Production Readiness: Instrumented agent systems with OpenTelemetry, structured JSON logging, LangSmith tracing, and token-level usage tracking for production monitoring and cost governance - Full-Stack Delivery: Built async Python FastAPI backends with real-time SSE streaming and Next.js/React frontends for agent interaction interfaces, including GitHub OAuth and webhook integrations

Education

• Liverpool John Moores University

  Master's degree, Data Science

• Vasantdada Patil Pratishthan’s College of Engineering and Visual Arts (VPPCOEVA)

  Bachelor of Engineering - BE, Computer Science

• Yeshiva University

  Master's degree, Artificial Intelligence

• International Institute of Information Technology Bangalore

  P.G. Diploma, Data Science