I’m a PhD student working on AI systems at the UC Berkeley Sky Computing Lab (previously RISELab), where I’m advised by Professor Ion Stoica and Professor Joseph Gonzalez.

I’m interested in how techniques from computer systems can make self-driving cars safer and more reliable. To that effect, I’ve co-designed 2 open-source software projects:

• ERDOS is a dataflow system for building self-driving car and robotics applications.
• Pylot is a modular self-driving car platform that runs on both real cars and in simulation.

Previously, I contributed to several successful open-source projects:

• Ray is a popular framework for developing distributed applications.
• Modin scales pandas workflows to terabytes.

I interned at Google Deepmind with the Pathways Team in 2023, where I focused on sparse computation and modularity in language models.

Publications

• Scalable Multi-Domain Adaptation of Language Models using Modular Experts
  Peter Schafhalter*, Shun Liao*, Yanqi Zhou, Chih-Kuan Yeh, Arun Kandoor, James Laudon.
  [website] [paper]
• Bandwidth Allocation for Cloud-Connected Autonomous Driving
  Peter Schafhalter*, Alexander Krentsel*, Joseph E. Gonzalez, Sylvia Ratnasamy, Scott Shenker, Ion Stoica. Under submission, conference paper.
• MoE-Lightning: High-Throughput MoE Inference on Memory-Constrained GPUs
  Shiyi Cao, Shu Liu, Tyler Griggs, Peter Schafhalter, Xiaoxuan Liu, Ying Sheng, Joseph E. Gonzalez, Matei Zaharia, Ion Stoica. ASPLOS 2025.
  [website] [paper] [code]
• Managing Bandwidth: The Key to Cloud-Assisted Autonomous Driving
  Alexander Krentsel*, Peter Schafhalter*, Joseph E. Gonzalez, Sylvia Ratnasamy, Scott Shenker, Ion Stoica.
  [paper]
• The Sky is the Limit: Cloud-Assisted Autonomous Driving via Service Tiers
  Alexander Krentsel, Peter Schafhalter, Joseph E. Gonzalez, Sylvia Ratnasamy, Scott Shenker, Ion Stoica. Compound AI Systems Workshop 2024.
  [extended abstract] [poster]
• Leveraging Cloud Computing to Make Autonomous Vehicles Safer
  Peter Schafhalter, Sukrit Kalra, Le Xu, Joseph E. Gonzalez, Ion Stoica. IROS 2023.
  [website] [paper] [slides] [poster]
• Context-Aware Streaming Perception in Dynamic Environments
  Gur-Eyal Sela, Ionel Gog, Justin Wong, Kumar Krishna Agrawal, Xiangxi Mo, Sukrit Kalra, Peter Schafhalter, Eric Leong, Xin Wang, Bharathan Balaji, Joseph Gonzalez, Ion Stoica.
  [paper] [code]
• D3: A Dynamic Deadline-Driven Approach for Building Autonomous Vehicles
  Ionel Gog, Sukrit Kalra, Peter Schafhalter, Joseph E. Gonzalez, Ion Stoica. EuroSys 2022.
  [website] [paper] [code] [docs | rust] [docs | python]
• Pylot: A Modular Platform for Exploring Latency-Accuracy Tradeoffs in Autonomous Vehicles
  Ionel Gog*, Sukrit Kalra*, Peter Schafhalter*, Matthew A. Wright, Joseph E. Gonzalez, Ion Stoica. ICRA 2021. Best Unmanned Aerial Vehicles paper award finalist.
  [website] [paper] [code] [docs] [video]

* These authors have made equal contributions.

Selected Talks

Towards Cloud-Driven Autonomous Vehicles. Sky 2024 Summer Retreat.
[slides]

Building Autonomous Vehicles: From the Road to the Sky. Sky 2022 Summer Retreat.
[slides]

D3: A Dynamic Deadline-Driven Approach for Building AVs. EuroSys 2022.

D3: A Dynamic Deadline-Driven Approach for Building AVs. RISE/Sky 2022 Winter Retreat.
[slides]

Scaling Emerging AI Applications with Ray. QCon.ai San Francisco 2019.

Service

Reviewer: ICRA 2024, IROS 2024, ICLR 2025.

Co-author of the UC Berkeley EECS Operating Systems Preliminary Exam Syllabus (2023).