Harshad Khadilkar

Harshad Khadilkar is an AI researcher whose work applies machine learning, optimization and reinforcement learning to transportation, supply chains and finance. He earned an aerospace BTech at IIT Bombay in 2009, an SM at MIT in 2011 and a PhD there in 2013. After IBM Research, he worked at TCS Research, where his group used reinforcement learning for supply-chain decisions; he then held research roles at Franklin Templeton and a data science role at Tata. He is a visiting associate professor at IIT Bombay, and his recent work includes railway scheduling, scholarly AI agents and question-generation evaluation. As of October 2026, he is Franklin Templeton's Director and Principal Research Scientist, working on reliable generative AI and AI-driven investment methods.

Now at: IIT Bombay, IIT Madras and Franklin Templeton
Badges: Bridge

Timeline

2009–2013

Degrees and airport models

He earns a BTech in aerospace engineering at IIT Bombay in 2009, an SM in aeronautics and astronautics at MIT in 2011, and a PhD in the same field at MIT in 2013.
August 2013–May 2015

IBM Research scientist

He works as a research scientist at IBM Research from August 2013 to May 2015.
May 2015–November 2023

Reinforcement learning at TCS

From May 2015 to November 2023, he works as a scientist and senior scientist at TCS Research. In 2022, he heads a TCS research group that applies reinforcement learning to move inventory efficiently through supply chains.
Since August 2019

A visiting role at IIT Bombay

Since August 2019, he serves as a visiting associate professor in IIT Bombay's Computer Science and Engineering department. He is also an associate research fellow at RBCDSAI, with research interests in stochastic control, machine learning for optimization, and air and rail traffic management.
November 2023–January 2026

Franklin and Tata leadership roles

At Franklin Templeton, he serves as Director and Principal Research Scientist from November 2023 through March 2025. In April 2025, he moves to Tata Sons as Data Science Lead, focusing on making generative AI more reliable and capable through January 2026.
2024

Deep reinforcement learning for rail

In 2024, he coauthors a deep reinforcement learning approach for routing and scheduling trains at railway stations with TCS Research collaborators.
2025

AI agents and question evaluation

In September 2025, he coauthors AEGIS, an agent that identifies proceedings papers by geographic region and can prepare nomination forms; its abstract reports 100% recall and 99.4% accuracy across 586 papers. In December 2025, he coauthors AURA-QG, a pipeline that scores question sets for answerability, non-redundancy, coverage, and structural entropy.
Since January 2026

Research leadership at Franklin Templeton

He returns to Franklin Templeton as Director and Principal Research Scientist in January 2026. As of October 2026, he is also a visiting associate professor at IIT Bombay and an associate research fellow at RBCDSAI; his work includes AI-driven investment methods and making generative AI more reliable and capable.