Efficient LLM Training and Inference

Developing efficient algorithms and systems for large language model training and inference, focusing on memory optimization, distributed computing, and hardware-aware acceleration techniques.

AI/Agentic Infra

Building scalable infrastructure for AI agents and autonomous systems, including multi-agent coordination, tool-use frameworks, and reliable agentic workflows.

Systems for Embodied AI

Designing efficient computing systems for embodied AI applications, including real-time perception, planning, and control on resource-constrained edge platforms.

Computing-in-Memory Architecture and Toolchains

Exploring novel processing-in-memory and computing-in-memory architectures, along with end-to-end toolchains for compilation, mapping, and performance evaluation.

Software-Hardware Co-design of Cryptography

Co-designing cryptographic algorithms and hardware accelerators for privacy-preserving computation, including homomorphic encryption, secure multi-party computation, and post-quantum cryptography.