Company: TCS
Role: Full Stack AI Architect
Project: DNB (Enterprise Banking)
About Project:
In this role, I led the architectural design and development of the e-Zenit AI Customer Onboarding platform, transforming a mission-critical enterprise banking solution into a fully Agentic AI ecosystem. The platform was built to create a single intelligent AI interface for the seamless onboarding of Norwegian, Non-Norwegian, institutional, and financial customers. The platform enables seamless onboarding of Norwegian, Non-Norwegian, institutional, and financial customers through a single intelligent AI interface, which automates end-to-end KYC/Kundekontroll, AML compliance, customer creation, and account setup workflows
Key Technical Works:
- Designed a multi-agent onboarding architecture utilizing LangChain and LangGraph to enable autonomous decision-making across customer onboarding workflows.
- Built Agentic AI workflows to handle KYC, AML validation, tax reporting, customer classification, and agreement management.
- Developed enterprise-grade RAG pipelines for the retrieval of banking regulations, compliance documents, and customer records using Azure Cosmos DB Vector Search and Azure Blob Storage.
- Implemented AI-driven customer onboarding across multiple channels (Netbank, Mobile Banking, internal CRM systems) using Zenit Kundefront as the central orchestration layer.
- Integrated core banking systems such as Felix (for customer data) and KVH (for CRM/sales intelligence) to provide real-time customer insights directly to the AI agents.
- Designed secure AI middleware using Azure AI Foundry to orchestrate private LLMs (Ollama) with a fallback to GPT-based models for complex reasoning tasks.
- Implemented compliance-aware AI guardrails to ensure strict adherence to banking regulations, auditability, and role-based access control (RBAC).
- Established automated LLMOps/MLOps pipelines for deployment and infrastructure management using GitHub Actions and Terraform (IaC).
- Integrated RAG evaluation frameworks (Ragas) to continuously validate retrieval accuracy and response reliability.
- Built comprehensive observability dashboards utilizing Splunk and Azure Application Insights to monitor system latency, cost, token usage, and overall health.
- Delivered a fully automated onboarding system that significantly reduced the need for manual intervention.
Skill Sets:
LangChain, LangGraph, Azure Cosmos DB Vector Search, Azure Blob Storage, Zenit Kundefront, Felix, KVH, Azure AI Foundry, Ollama, GPT Models, GitHub Actions, Terraform (IaC), Ragas, Splunk, and Azure Application Insights