Niranjan VSKS

I'm a senior agentic AI engineer. I ship into enterprise environments end to end: cold-start discovery, architecture, implementation, and production deployment.

Seven years in, across data science, GenAI, and agentic AI, the lesson that stuck is that the model is rarely the hard part. The hard part is getting into a customer's real environment, understanding the actual problem quickly, and shipping production-grade systems that integrate seamlessly.

• Mphasis (2019-2022): built fraud-detection and risk-scoring models for financial-services clients using tree-based ensembles (XGBoost, LightGBM, Random Forest). Where I learned to ship models against real data.
• HPE (2022-2023): architected the migration of an enterprise conversational system from rule-based FAQ to LLM-assisted RAG over ~1,700 internal documents, cutting repeat escalations ~40%. Where I learned retrieval at enterprise scale.
• Coforge (2024-present): architected and shipped an agentic quality-engineering platform from feasibility through production rollout to 17 enterprise QA teams at peak adoption, across customer-managed AWS, Azure, and GCP. Where I learned to deliver agentic systems inside real customer constraints.

Alongside the day job I build and ship my own products: Loop Copilot (live, Fortune 500 pilot), a voice-first financial copilot for India's gig workforce, and an autonomous multi-asset portfolio agent.

Architect first, ship fast, validate with users. I make architectural calls under real customer constraints and I own the outcome end to end.

Agentic AI on real customer data, with real consequences, is where the next decade of AI engineering happens. That is the room I want to be in.