Chiranjoy Chowdhuri

Chiranjoy Chowdhuri runs AI and data globally for Dr. Reddy's, which makes his the most regulated mandate in this cohort — models operating inside drug development, regulatory workflows, supply chain and commercial functions, where a wrong output carries compliance consequences on top of P&L ones. Since October 2024 the portfolio has spanned GenAI copilots, decision support for financial planning, intelligent document processing for regulatory submissions, demand forecasting for supply planning, and omnichannel marketing platforms, plus a new AI Labs being built in Hyderabad to work directly on LLM, RAG and agentic systems. His public writing is the tell on depth: a three-part series on hidden failure modes in production GenAI — signal contamination, silent reasoning skips between agents, judge anchoring and reward hacking — reads like an SRE runbook, down to checksum-at-handoff specs and retry-rate alert thresholds. Almost nobody at his grade publishes instrumentation detail; he treats it as the job.

The route there explains the engineering reflex. An M.Tech in industrial engineering and operations research from IIT Bombay — his thesis produced a scheduling algorithm 10x faster than CPLEX at 99.8% optimality — then six years in McKinsey's advanced analytics practice advising 20-plus Fortune 500 clients across the US, EMEA and APAC. The operator phase since has the rare property of building the same muscle in four industries: JPMorgan Chase consumer banking (demand forecasting, Monte Carlo resiliency planning), Mahindra's group strategy office (a unified customer data lake and ML personalisation across auto, finance and farm businesses), Pidilite (a Data and Analytics CoE from scratch — outlet-level recommendation engines, trade-promotion optimisation, stockout prediction), and Mankind Pharma (Azure Databricks lakehouse, o9 demand sensing across 1,000+ SKUs). AIM's AI100 named him among India's most influential AI leaders in 2022. The archetype: the engineer-consultant who has stood up the enterprise AI playbook four times and now runs it where the stakes are regulatory.