- Academics
-
Research
-
Centres
- E-Health Research Centre (EHRC)
- Machine Intelligence & Robotics CoE (MINRO)
- Centre for IT & Public Policy (CITAPP)
- Cognitive Computing CoE (CCC)
- Centre for Accessibility in the Global South (CAGS)
- COMET Tech Innovation Hub (NM-ICPS)
- IIITB Innovation Centre
- Modular Open-Source Identity Platform (MOSIP)
- Centre for Open Societal Systems (COSS)
- Centre for Digital Public Infrastructure | CDPI
-
Labs
- Surgical and Assistive Robotics Lab
- Scalable Data Science and AI Lab
- Graphics-Visualization-Computing-Lab
- Web Science Lab
- Multimodal Perception Lab
- Software Engineering and Analysis Lab
- High Density Electronic Systems Lab
- Networking and Communication Lab
- Remote Sensing, GIS and Spatial Computing Lab
- Indian Knowledge System (IKS) Lab
- Smart City Lab
- Ascend Studio
- Radar Sensing Lab
- CSSMP
- Advanced Wireless Communications Lab
- Speech Lab
- Connected Devices and Wearables Lab
- Outreach
- Publications
- Policy
- IIIT-B Press
-
Centres
- Placements
- Campus Life
- Media
- People
- About Us
Congratulations to Shashank Tippanavar (5th-year Integrated M.Tech) and Vivek Yadav (Technology Officer and Adjunct Faculty) on receiving the Runner-Up Award for Best Paper in the Research Track at the 6th International Conference on AI-ML Systems for their paper titled "AdapterSlots: Warp-Aligned Multi-LoRA Serving via WAR-Gated Kernel Promotion." The conference was held from October 6–9, 2026 at the Politecnico di Milano, Lecco Campus, Italy.
About the Paper: The team developed AdapterSlots, a scheduling system designed to improve the efficiency of multi-tenant Large Language Model (LLM) inference with concurrent LoRA adapters. The system uses warp-aligned request batching and adaptive GPU kernel selection to reduce memory-bandwidth overhead and accelerate inference. It achieves up to 1.5× higher throughput compared to vLLM serving.