Rahul Kumar Jha

Architect and deploy enterprise-grade Agentic RAG services, multi-modal GenAI pipelines, and production LLM microservices.

• Developed an Agentic RAG system on Azure (AI Foundry, OpenAI, LangChain) for brand compliance and marketing intelligence, cutting search latency by 45%.
• Engineered multi-modal Generative AI pipelines combining OpenCV and LLM-driven structured extraction across 10,000+ daily assets.
• Built low-latency FastAPI and Docker microservices for production LLM serving with prompt caching and context optimization.
• Implemented MLflow LLM tracing and automated evaluation pipelines, optimizing container inference throughput by 30%.

Engineered internal Generative AI and NLP solutions, semantic retrieval pipelines, and automated fraud-scoring logic for global banking clients.

• Engineered internal Generative AI and NLP solutions for banking clients, automating extraction and summarization of complex financial transaction logs.
• Developed semantic search and retrieval pipelines utilizing vector embeddings and similarity indexing to surface suspicious transaction narratives.
• Operationalized automated fraud-scoring logic, verifying LLM outputs against ground-truth account records to improve anomaly detection precision.

Engineered automated NLP text extraction pipelines, intelligent auditing services, and data ingestion architectures on AWS and MySQL.

• Engineered automated NLP text extraction and data ingestion pipelines utilizing AWS (S3, CloudWatch, ECS) and MySQL, reducing report latency by 55%.
• Constructed intelligent auditing services and parsing engines, replacing manual auditing workflows with automated text validation.
• Designed modular validation scripts and automated testing routines, accelerating client AI automation rollout timelines by 25%.

Built backend data processing workflows and integrated machine learning endpoints for an AI-powered talent intelligence platform.

• Engineered data ingestion pipelines handling thousands of unstructured candidate profiles and resumes daily.
• Integrated RESTful APIs and ML model inference endpoints for automated candidate screening and entity extraction.
• Optimized complex SQL queries and relational schemas in PostgreSQL, achieving a 35% reduction in search query latency.