Sayan Ranu

DS Chair Professor in AI · IIT Delhi

Graph ML
Algorithm Discovery
Agentic Systems
Self-Supervised Learning
Trustworthy AI
Sayan Ranu

I hold the position of DS Chair Professor in AI at IIT Delhi, appointed jointly in the Department of Computer Science and Engineering and the Yardi School of AI. My group works on machine learning for structured data, and increasingly on getting models to discover and act on their own. For a long time our main focus has been graphs: the networks, molecules, and relational systems behind things like road maps, recommendation engines, and new materials. We design new architectures, try to work out what these models actually learn, and find ways to make them fast and reliable at scale.

More recently we have been chasing a bigger question: can AI agents reason, plan and propose novel solutions to unknown problems? That has led us into automated algorithm discovery, where we use LLMs to synthesize algorithms for hard combinatorial problems. It has also drawn us toward agentic systems that plan and reason over many steps, and self-supervised learning such as JEPA. Across all of it we care a lot about trustworthy AI: models that are interpretable, fair, robust, and willing to admit when they are unsure.

Our papers regularly land at venues like NeurIPS, ICML, ICLR, and KDD. This also doubles as our unofficial travel program, and the group photo album has picked up a suspicious number of mountains and beaches over the years. You can find the full list of publications on my publications page.

Of course, none of this happens without the group. Behind these papers is a small army of PhD students, dual-degree and B.Tech students, and alumni who have since scattered to PhD programs and labs around the world. They do the real work, pick up the occasional award, and keep me honest in group meetings. If you are curious who they are and where they have ended up, come meet the group.

I am always looking for PhD students who get excited about hard problems and the possibility of traveling to exotic places! If any of this sounds like your kind of thing, do consider applying to PhD programs at CSE and ScAI, IIT Delhi. We work hard, but we have a good time doing it.

Our research has been generously supported by industry partners including Google, Mastercard AI Garage, Fujitsu Research, Flipkart, Incept Labs, and Dolby, among others.

Internships: I am NOT offering any short-term projects. My apologies if I don't respond to your email.

News

2026–27 Program Co-Chair of the Datasets & Benchmarks Track at KDD 2027, and Senior Area Chair at NeurIPS 2026.

May 2026 Three papers at ICML 2026, on model merging for GNNs, graph condensation, and solubility prediction on molecular graphs.

Apr 2026 Is graph unlearning ready for practice? Read our ICLR 2026 paper for the answers.

Jan 2026 Appointed DS Chair Professor in AI at IIT Delhi (2026–2028).

Dec 2025 · NeurIPS Oral GnnXemplar was selected for an Oral at NeurIPS 2025: the top 0.36% of submissions, and the only Oral from India this edition. Joint work with Fujitsu Research on turning GNN predictions into natural-language rules.

Dec 2025 Recognized as a Top-10% Area Chair at NeurIPS 2025.

May 2025 A hat-trick at KDD 2025, all on machine learning for graphs.

2024–25 Received the Teaching Excellence Award, IIT Delhi.