Skip to main content
ahad.

AI Engineer

I build stateful agent graphs with LangGraph, MCP tool servers with FastAPI, and hybrid retrieval pipelines on Azure AI Search. 5,000+ enterprise documents processed. Currently at Capgemini.

Experience

Capgemini

Dec 2024 – Present

India

AI Engineer

  • Architect stateful LangGraph StateGraph workflows with PostgresSaver checkpointing — every agent run is resumable and has a full audit trail
  • Built IT access provisioning automation with human-in-the-loop approval gates using LangGraph interrupt/resume — replaced a manual ticket queue
  • Implement tool-calling patterns for autonomous database querying using MCP; agents discover and call tools without hardcoded schemas
  • Build data ingestion pipelines: PDF/DOCX parsing, table extraction, recursive chunking with overlap for retrieval accuracy
  • Develop FastAPI async backends with streaming responses (SSE) for real-time agent state telemetry to the browser
  • Implement hybrid retrieval on Azure AI Search: BM25 keyword + dense vector + Reciprocal Rank Fusion (1 / (60 + rank)) + semantic reranking
  • Design database connectivity layers so agents query SQL databases via natural language — no ORM boilerplate in agent code
  • Apply vector search and embeddings for semantic retrieval and context grounding across 5,000+ document corpora

Currently Exploring

Agent-to-Agent (A2A) Protocols

Decentralized inter-agent negotiation and standardized communication schemas across heterogeneous agent networks

Structured Output Guarantees

100% deterministic JSON generation via Outlines & SGLang finite-state machine token masking

Edge SLM & Quantization

Deploying 3B-7B parameter models (Qwen 2.5 & DeepSeek R1 distilled) on local Apple Silicon & edge hardware with 0 API cost

Hybrid Graph RAG

Combining Knowledge Graphs (Neo4j / GQL) with dense vector search for multi-hop reasoning across 10K+ documents

Core Expertise

Python Backend

FastAPI, async patterns, REST APIs, microservices architecture, and scalable data pipelines.

Agent Systems & RAG

Retrieval-augmented generation, vector search, tool calling, function calling, LangGraph StateGraph orchestration, MCP tool discovery, and data pre-processing pipelines.

Cloud & DevOps

Azure AI services, Docker containerization, CI/CD pipelines, and production deployments.

Technical Skills

Agentic Orchestration

LangGraphLangChainStateGraphCheckpointingHITLinterrupt/resumeConditional Edges

Protocols & Interfaces

FastAPIMCP (Model Context Protocol)SSEREST APIsTool CallingWebSockets

Vector Search & RAG

Azure AI SearchBM25RRFDense VectorChromaDBMilvusEmbeddingsReranking

Languages & Infra

PythonSQLTypeScriptDockerAzurePostgreSQLGit

Education

B.Tech in Information Technology

GL Bajaj Institute of Technology & Management

2020 – 2024 · Greater Noida, Uttar Pradesh