A*STAR CDF | Algorithm-System Co-Design for Efficient and IP-Protected LLMs

PI. Funded by A*STAR Career Development Fund (CDF). 31 awardees selected from 194 applicants. S$250,000.

Project Overview

Title: Algorithm-System Co-Design for Efficient and IP-Protected LLMs: From Model Optimization to Cluster Deployment

Role: Principal Investigator (PI)

Funding: S$250,000

Funder: A*STAR Career Development Fund (CDF)

Selection: 31 recipients selected from 194 applicants

Objectives

This project tackles two intertwined challenges in deploying large language models (LLMs) at scale:

  1. Efficiency — reducing inference latency and memory footprint via algorithm-hardware co-design, covering sparse MoE routing, speculative decoding, and KV cache compression.

  2. IP Protection — embedding and verifying model ownership at the cluster level, so that LLM capabilities can be licensed and traced even after fine-tuning or distillation.

The project spans the full stack: from per-layer optimization algorithms, through runtime scheduling on GPU/NPU clusters, to end-to-end deployment pipelines.

Affiliation

Agency for Science, Technology and Research (A*STAR), Singapore
Centre for Frontier AI Research (CFAR)