PRIYANSH KEDIA

I’m an Applied AI Engineer with 5 years of experience building, deploying, and owning production ML and AI systems at scale. I’ve served 15M+ users by taking ambiguous business problems and unstructured data and shipping a production intelligence layer with full end-to-end ownership—from research and deployment to monitoring and incident response.

At CRED, I was the primary AI engineer and DS owner across user-facing products like CRED Money, CRED Card Management, and Guardian. I architected CMAAS, a transaction intelligence platform that converts raw financial inputs (SMS, email, account aggregators, and statements) into structured enriched transactions using custom Transformer models, NER, and multi-signal fusion, alongside real-time and offline monitoring frameworks and continuous improvement loops; I also extended intelligence across multiple financial instruments and built agentic AI workflows with a multi-LLM stack and hybrid RAG with citation grounding. I focus heavily on reliability engineering, building LLM-evaluation pipelines and autonomous reliability frameworks (including human-in-the-loop validation) so models stay trustworthy as data patterns drift; I’ve owned the full ML lifecycle across ~19 microservices and ~30 batch jobs, with observability, alerting, and cost/latency optimization to keep production performance strong.