Alvin Cheung

Room 785, Soda Hall Follow @alvinkcheung Mailing address 387 Soda Hall Berkeley, CA 94720-1776

I am an associate professor in the Computer Science Division at UC Berkeley EECS. I am a member of the Data Systems and Foundations group, Programming Systems group, Sky Lab, SLICE Lab, and a faculty affiliate in the Berkeley Institute for Data Science. I also serve as an advisor to the Data Science Discovery Program, and technical advisor to several companies.

My research interests include data management, programming languages, and building software systems. My group aims to help end users, from data scientists to coding experts, to easily extract insights from large amounts of data. We also develop techniques and tools that make it easy to build large-scale, efficient, and manageable data processing pipelines, where such pipelines range from traditional computer science applications (web, cloud) to emerging application domains in data science (physical sciences, healthcare, social sciences) and beyond.

Some research themes:

• Verified lifting is a new technique for inferring properties of programs. We have applied this to database applications (QBS), stencil computations (STNG), programmable switches (Domino), parallel data processing frameworks (Casper), GPUs (Dexter and Rake), and CRDTs (Katara). Code generated using our techniques are now deployed at Adobe and Google.
• Designing new data processing and programming languages techniques to build and optimize data management and machine learning systems for emerging applications: Cosette (reasoning about data transformations from grading homework assignments to building query optimizers), Hyperloop (improving the performance of database-backed applications), Spatialyze (data management system for large-scale geospatial video datasets), and OSD (adaptive speculative decoding).
• Improving end user (such as data scientists) data programming experience with new user interfaces and code generators across different domains, using techniques from machine learning, program synthesis, and human-computer interaction: PlotCoder, Falx, AST-T5, and SAFIM.

If you are a Berkeley student interested in doing research in these areas and have done well in CS186, CS164, or CS182 please send me an email mentioning which class(es) you have taken, and include your resume and unofficial transcript.

I regularly teach the undergraduate data management class at Berkeley. Check out the recorded videos if you are interested.

We thank the NSF, DOE, ONR, ARO, Adobe, Google, Intel, Meta, and VMware for their generous support of our group's research, and transferring our technology to real-world products. We are also grateful for the early career awards and other recognitions from the data management and programming systems research communities.

I was earlier on the CSE faculty at the University of Washington and an affiliate in the UW eScience Institute, and before that a graduate student in the MIT database group and the computer-aided programming group, working with Professors Sam Madden and Armando Solar-Lezama. I worked on tools that make use of programming language techniques to improve application performance.

What's new?

• [10/24] Honored to receive the Dahl-Nygaard Junior Prize this year!
• [7/24] We will be presenting 3 papers at ICML this year: Lily's work on continuously updating draft models to improve speculative decoding performance, and Linyuan's work on making LLMs code-aware for coding tasks along with a new benchmark for evaluation.
• [7/24] We just released Mélange for deciding how to allocate GPUs for LLM workloads. Check it out!
• [5/24] We released a new benchmark for LLM-based coding tasks. Unlike traditional code completion type tasks, our benchmark requires the model to understand both the code that comes before and after the part to be generated. Take a look at our leaderboard.
• [8/23] Excited to receive the VLDB Early Career Research Contributions Award this year. Thank you mentors, colleagues, and students for your support and guidance!

Teaching

• CS 186: Introduction to Database Systems
• CS 263: Design of Programming Languages (Fall 19, Fall 21)
• CSE 344: Introduction to Data Management (Winter 15, Fall 16, Winter 17)
• CSE 401: Hack Your Language! (Winter 16)
• CSE 414: Introduction to Database Systems (Spring 18)
• CSE 544: Principles of Data Management (Fall 15)
• CSE 501: Principles and Applications of Program Analysis (Spring 15)
• CSE P 544: Database Management Systems (Fall 17)

Students who I humbly learn from

• Sahil Bhatia (co-advised with Sanjit Seshia)
• Linyuan Gong
• Mick Kittivorawong
• Shadaj Laddad (co-advised with Joe Hellerstein)
• Lily Liu

Graduated students

• Gabriel Matute (co-advised with Sarah Chasins) (MS 2024)
• Maaz Ahmad (PhD 2022, Adobe Research)
• Chenglong Wang (co-advised with Ras Bodik) (PhD 2021, Microsoft Research)
• Cong Yan (PhD 2020, Microsoft Research)
• Brandon Haynes (co-advised with Magda Balazinska) (PhD 2020, Microsoft Jim Gray Systems Lab)
• Srini Iyer (co-advised with Luke Zettlemoyer) (PhD 2020, Facebook AI Research)
• Guna Prasaad (co-advised with Dan Suciu) (MS 2020, Facebook)
• Tomer Kaftan (co-advised with Magda Balazinska) (MS 2018, Google Brain)

Papers

• Flo: a Semantic Foundation for Progressive Stream Processing
  Shadaj Laddad, Alvin Cheung, Joseph M. Hellerstein, Mae Milano
  POPL 2025
• Verified Code Transpilation with LLMs
  Sahil Bhatia, Jie Qiu, Niranjan Hasabnis, Sanjit A. Seshia, Alvin Cheung
  NeurIPS 2024
• Inferring Visualization Intent from Conversation
  Haotian Li, Nithin Chalapathi, Huamin Qu, Alvin Cheung and Aditya G. Parameswaran
  CIKM 2024
• Tenspiler: A Verified Lifting-Based Compiler for Tensor Operations
  Jie Qiu, Colin Cai, Sahil Bhatia, Niranjan Hasabnis, Sanjit Seshia, Alvin Cheung
  ECOOP 2024
• QED: A Powerful Query Equivalence Decider for SQL
  Shuxian Wang, Sicheng Pan, Alvin Cheung
  VLDB 2024
• Evaluation of LLMs on Syntax-Aware Code Fill-in-the-Middle Tasks
  Linyuan Gong, Sida Wang, Mostafa Elhoushi, Alvin Cheung
  ICML 2024
• AST-T5: Structure-Aware Pretraining for Code Generation and Understanding
  Linyuan Gong, Mostafa Elhoushi, Alvin Cheung
  ICML 2024
• Online Speculative Decoding
  Xiaoxuan Liu, Lanxiang Hu, Peter Bailis, Alvin Cheung, Zhijie Deng, Ion Stoica, Hao Zhang
  ICML 2024
• ADELT: Transpilation Between Deep Learning Frameworks
  Linyuan Gong, Jiayi Wang, Alvin Cheung
  IJCAI 2024
• SlimFit: Memory-Efficient Fine-Tuning of Transformer-based Models Using Training Dynamics
  Arash Ardakani, Altan Haan, Shangyin Tan, Doru Thom Popovici, Alvin Cheung, Costin Iancu, Koushik Sen
  NAACL 2024
• Syntactic Code Search with Sequence-to-Tree Matching
  Gabriel Matute, Wode Ni, Titus Barik, Alvin Cheung, Sarah E. Chasins
  PLDI 2024
• Spatialyze: A Geospatial Video Analytics System with Spatial-Aware Optimizations
  Chanwut Kittivorawong, Yongming Ge, Yousef Helal, Alvin Cheung
  VLDB 2024
• Towards Auto-Generated Data Systems
  Alvin Cheung, Maaz Bin Safeer Ahmad, Brandon Haynes, Chanwut Kittivorawong, Shadaj Laddad, Xiaoxuan Liu, Chenglong Wang, Cong Yan
  VLDB 2023
• Building Code Transpilers for Domain-Specific Languages Using Program Synthesis
  Sahil Bhatia, Sumer Kohli, Sanjit Seshia, Alvin Cheung
  ECOOP 2023
• Model-Generated Pretraining Signals Improves Zero-Shot Generalization of Text-to-Text Transformers
  Linyuan Gong, Chenyan Xiong, Xiaodong Liu, Payal Bajaj, Yiqing Xie, Alvin Cheung, Jianfeng Gao, Xia Song
  ACL 2023
• Optimizing Stateful Dataflow with Local Rewrites
  Shadaj Laddad, Conor Power, Tyler Hou, Alvin Cheung, Joseph M. Hellerstein
  EGRAPHS 2023
• Leveraging Application Data Constraints to Optimize Database-Backed Web Applications
  Lily Liu, Shuxian Wang, Mengzhu Sun, Sicheng Pan, Ge Li, Siddharth Jha, Cong Yan, Junwen Yang, Shan Lu, Alvin Cheung
  VLDB 2023
• Keep CALM and CRDT On
  Conor Power, Mae Milano, Shadaj Laddad, Alvin Cheung, Joseph Hellerstein, Natacha Crooks
  VLDB 2023
• Synthesizing CRDTs from Sequential Data Types with Verified Lifting
  Shadaj Laddad, Conor Power, Mae Milano, Alvin Cheung, Joseph M. Hellerstein
  OOPSLA 2022
• GACT: Activation Compressed Training for Generic Network Architectures
  Lily Liu, Lianmin Zheng, Dequan Wang, Yukuo Cen, Weize Chen, Xu Han, Jianfei Chen, Zhiyuan Liu, Jie Tang, Joey Gonzalez, Michael Mahoney, Alvin Cheung
  ICML 2022
• Synthesizing Analytical SQL Queries from Computation Demonstration
  Xiangyu Zhou, Ras Bodik, Alvin Cheung, Chenglong Wang
  PLDI 2022, distinguished paper award
• Vector Instruction Selection for Digital Signal Processors Using Program Synthesis
  Maaz Bin Safeer Ahmad, Andrew Adams, Shoaib Kamil, Alvin Cheung, Alexander J. Root
  ASPLOS 2022
• Change in Software Ecosystems: Social Challenges of Automating Upgrades
  Gabriel Matute, Alvin Cheung, Sarah E. Chasins
  PLATEAU 2021
• Demonstration of Apperception: A Database Management System for Geospatial Video Data
  Vanessa Lin, Frank Ge, Brandon Haynes, Maureen Daum, Alvin Cheung, Magdalena Balazinska
  VLDB 2021
• PlotCoder: Hierarchical Decoding for Synthesizing Visualization Code in Programmatic Context
  Xinyun Chen, Linyuan Gong, Alvin Cheung, Dawn Song
  ACL 2021
• VSS: A Storage System for Video Analytics
  Brandon Haynes, Maureen Daum, Dong He, Amrita Mazumdar, Magdalena Balazinska, Alvin Cheung, Luis Ceze
  SIGMOD 2021
• Falx: Synthesis-powered Visualization Authoring
  Chenglong Wang, Yu Feng, Rastislav Bodik, Isil Dillig, Alvin Cheung, Amy J. Ko
  CHI 2021, best paper award
• New Directions in Cloud Programming
  Alvin Cheung, Natacha Crooks, Joseph Hellerstein, Mae Milano
  CIDR 2021
• Handling Highly Contended OLTP Workloads using Fast Dynamic Partitioning
  Guna Prasaad, Alvin Cheung, Dan Suciu
  SIGMOD 2020
• Demonstration of Chestnut: An In-memory Data Layout Designer for Database Applications
  Mingwei Samuel, Cong Yan, Alvin Cheung
  SIGMOD 2020
• Testing Query Execution Engines with Mutations
  Xinyue Chen, Chenglong Wang, Alvin Cheung
  DBTest 2020
• Managing Data Constraints in Database-Backed Web Applications
  Junwen Yang, Utsav Sethi, Cong Yan, Shan Lu, Alvin Cheung
  ICSE 2020
• Visualization by Example
  Chenglong Wang, Yu Feng, Rastislav Bodik, Alvin Cheung, Isil Dillig
  POPL 2020
• LightDB++ : A DBMS for the Visual World
  Bandon Haynes, Maureen Daum, Amrita Mazumdar, Magdalena Balazinska, Alvin Cheung, Luis Ceze
  CIDR 2020
• View-Driven Optimization of Database-Backed Web Applications
  Cong Yan, Alvin Cheung, Junwen Yang, Shan Lu
  CIDR 2020
• Automatically Translating Image Processing Libraries to Halide
  Maaz Bin Safeer Ahmad, Jonathan Ragan-Kelley, Alvin Cheung, Shoaib Kamil
  SIGGRAPH ASIA 2019
• Perceptual Compression for Video Storage and Processing Systems
  Amrita Mazumdar, Brandon Haynes, Magda Balazinska, Luis Ceze, Alvin Cheung, Mark Oskin
  SoCC 2019, best poster award
• Learning Programmatic Idioms for Scalable Semantic Parser
  Srini Iyer, Alvin Cheung, Luke Zettlemoyer
  EMNLP 2019
• Generating Application-Specific Data Layouts for In-Memory Databases
  Cong Yan, Alvin Cheung
  VLDB 2019
• Visual Road: A Video Data Management Benchmark
  Brandon Haynes, Amrita Mazumdar, Magdalena Balazinska, Luis Ceze, Alvin Cheung
  SIGMOD 2019
• View-Centric Performance Optimization for Database-Backed Web Applications
  Junwen Yang, Cong Yan, Chengcheng Wan, Shan Lu, Alvin Cheung
  ICSE 2019, distinguished paper award

Dissertation

• Rethinking the Application-Database Interface
  Alvin Cheung, MIT George M. Sprowls Award for best computer science dissertation
  

Recent Awards

• Dahl-Nygaard Junior Prize (2024)
• VLDB Early Career Research Contribution Award (2023)
• PLDI Distinguished Paper Award (2022)
• Army Research Office Early Career Program Award (2022)
• Office of Naval Research Young Investigator Award (2021)
• CHI Best Paper Award (2021)
• Intel Outstanding Researcher Award (2021)
• IEEE TCDE Rising Star Award (2020)
• VMware Early Career Faculty Award (2020)
• PECASE Award (2019)
• ICSE Distinguished Paper Award (2019)
• Sloan Fellowship (2019)
• Google Faculty Award (2018)
• SIGMOD Best Demo Award (2017)
• NSF CAREER Award (2017)
• Honorable mention, SIGMOD Jim Gray Dissertation Award (2016)
• US Department of Energy Early Career Research Award (2016)
• MIT George M. Sprowls Award for best computer science dissertation (2016)
• Google Faculty Award (2015)
• Intel PhD Fellowship
• NSF Graduate Fellowship
• NDSEG Fellowship
• Zipf Fellow, Council on Library and Information Resources

Service

• 2026:
  

  SIGMOD

  Associate Editor

• 2025:
  

  PVLDB	Associate Editor
  OOPSLA	Program Committee

• 2024:
  

  ECOOP

  Program Committee

• 2023:
  

  NeurIPS

  Program Committee

• 2022:
  

  POPL	Program Committee
  SIGMOD	Program Committee
  SPLASH	Program Committee

• 2021:
  

  SIGMOD	Program Committee
  ICDE	PhD Symposium Co-Chair
  VLDB Demo	Program Committee