Zeenhasn CG I

Architected and delivered production-grade GenAI applications using Python, Azure OpenAI, AWS Bedrock, and OpenAI APIs, building scalable RAG and agentic AI solutions integrated with cloud-native backend services. Implemented end-to-end RAG architectures (document ingestion/parsing/chunking, metadata enrichment, embeddings, vector indexing, retrieval, reranking, context assembly, prompt orchestration, inference, and response validation). Developed vector search/semantic retrieval layers using FAISS, pgvector, Pinecone, and Chroma, including metadata filtering and retrieval tuning. Built reusable FastAPI/REST microservices for LLM inference, embeddings, retrieval, evaluation, and agent capabilities with async processing, validation, authentication, caching, and structured observability. Created agent workflows using LangChain/LangGraph with tool calling, routing, state management, validation, and human-in-the-loop checkpoints. Established LLM evaluation frameworks (retrieval relevance,

Architected and delivered production-grade GenAI applications using Python and managed LLM services (Azure OpenAI, AWS Bedrock, OpenAI APIs, Hugging Face). Built scalable RAG and agentic AI/AI workflows with ingestion, parsing/chunking, embeddings, vector indexing and semantic retrieval, reranking, context assembly, prompt orchestration, validation, and guardrails. Developed reusable FastAPI/REST microservices for LLM inference, embeddings, retrieval, evaluation, and agent capabilities using async processing, auth, caching, structured observability, and robust error handling. Engineered LLM and RAG evaluation frameworks (retrieval relevance, groundedness, hallucination detection, prompt regression) and automated regression tests via CI/CD quality gates. Implemented ML lifecycle management using MLflow (tracking, versioning, evaluation artifacts, controlled promotion), plus monitoring and drift-aware reliability practices. Built enterprise data pipelines using Spark/Databricks, Kafka, A