Projects

• 
  MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs
  Overlap CPU and GPU computation with I/O to maximize utilization for offline, batch-processing LLM workloads.
  Shiyi Cao, Shu Liu, Tyler Griggs, Peter Schafhalter, Xiaoxuan Liu, Ying Sheng, Joseph E. Gonzalez, Matei Zaharia, Ion Stoica. ASPLOS 2024.
  [Paper] [Code]

---

• 
  Scalable Multi-Domain Adaptation of Language Models using Modular Experts
  Train and compose modular experts to augment the capabilities of language models.
  Peter Schafhalter*, Shun Liao*, Yanqi Zhou, Chih-Kuan Yeh, Arun Kandoor, James Laudon.
  [Paper]

---

• 
  Leveraging Cloud Computing to Make Autonomous Vehicles Safer
  Use the cloud to improve autonomous driving safety.
  Peter Schafhalter, Sukrit Kalra, Le Xu, Joseph E. Gonzalez, Ion Stoica. IROS 2023.
  [Paper] [Slides] [Poster]

---

• 
  D3: A Dynamic Deadline-Driven Approach for Building Autonomous Vehicles
  Adjust response times and handle missed deadlines to maximize application-wide accuracy.
  Ionel Gog, Sukrit Kalra, Peter Schafhalter, Joseph E. Gonzalez, Ion Stoica. EuroSys 2022.
  [Paper] [Code] [Docs | Rust] [Docs | Python]

---

• 
  Pylot: A Modular Platform for Exploring Latency-Accuracy Tradeoffs in Autonomous Vehicles
  Rapidly test self-driving car models and algorithms.
  Ionel Gog*, Sukrit Kalra*, Peter Schafhalter*, Matthew A. Wright, Joseph E. Gonzalez, Ion Stoica. ICRA 2021. Best Unmanned Aerial Vehicles paper award finalist.
  [Paper] [Code] [Docs] [Video]

*These authors have made equal contributions.