At Maveric Systems Limited, I led architecture and delivery of a production multi-agent AML investigation platform. Its LangGraph and AutoGen workflow processes real-time Kafka alerts and generates SAR reports for regulatory submission.
I built a banking RAG system on AWS Bedrock with MongoDB Atlas Vector Store, adding PII redaction, tracing and schema-validated outputs for compliance. I also architected PRISM, which turns regulatory PDFs into executable DMN/DRL rules through an asynchronous LLM workflow.
At Maveric Systems Limited, I built CodeLens, a microservices AI code-analysis platform using FastAPI agents and Tree-sitter AST parsing. I also shipped an MCP server that exposes banking tools through HTTP, WebSocket and STDIO.
At Capgemini Technology Services, I designed and shipped distributed RAG applications and production model-serving pipelines for finance, healthcare and cybersecurity clients. My work included adapting LLMs with QLoRA and LoRA, applying RLHF, and deploying workloads across AWS, GCP and Azure.
