Mihai Budiu's Web Page

About myself

Mihai Budiu Cell: (412) 216 8430 mbudiu (at) gmail mbudiu (at) feldera.com CV mihaibudiu.github.io github

2023-current: chief scientist, Feldera Inc. , an early stage start-up in the Bay Area, California. 2016-2023: senior researcher at VMware Research, in Palo Alto, California. 2014-2016: software engineer at Barefoot Networks, a start-up in Palo Alto, California. 2004-2014: senior researcher at the Microsoft Research Silicon Valley Lab. 2004: Ph.D. in computer science from Carnegie Mellon University. Engineering degree from Politehnica University of Bucharest. During the years I have participated in the design and construction of several complex software artifacts. I like building systems based on strong theoretical foundations. My research has spanned several broad areas of computer science: Databases Software-defined networking and P4 Cloud computing infrastructure: distributed computation platforms, monitoring, debugging, and performance analysis of large scale clusters Big data analysis and visualization Large scale machine learning Computer security Computer architecture Programming language and compiler design and implementation Reconfigurable hardware and hardware synthesis

Awards

Projects

• Incremental view maintenance (2021-today): DBSP (Database stream processing), a formal model and its implementation, and a SQL compiler for a streaming and incremental query engine.
• Differential Datalog, a fully incremental programming language based on Datalog (2018-2022). DDlog is used in production at VMware in the NSX network controller and as the data analytics engine of VMware Skyline.
• Nerpa: a system for programming software-defined networks (2017-2019; 2021-2023).
• Hillview, a spreadsheet for big data (2016-2021).
• The P4 programming language design and its reference implementation (2014-2023). I have been a co-chair of the P4 language design working group between 2018-2023.
• Compiling P4 to eBPF (2016-2021).
• The CosmosDB Azure Cloud database (at the time named DocumentDB) (2012-2014). I have worked on the design and implementation of the JSON-specific NoSQL query engine (language and runtime).
• The Dryad and DryadLINQ software stack (2005-2014). These are available under an Apache2 license on GitHub.
  • Dryad was the distributed execution engine of the Bing Cosmos analytics engine, deployed on tens of thousands of machines at Microsoft since 2006. The Dryad paper has received the EuroSys 2017 Test of Time award.
  • DryadLINQ is a compiler from .Net LINQ to Dryad. A commercial version of this compiler was shipped in 2011 by Microsoft in beta under the name LINQ to HPC but was afterwards discontinued. The DryadLINQ paper won the 2008 OSDI best paper award, and the ACM SIGOPS 2018 Hall of Fame award.
  • Daphne, DryadLINQ JobBrowser and debugger. Techniques introduced by this tool are used internally in Microsoft in the ScopeStudio debugger (the tool used for programming the Bing analytics engine).
• Cloud infrastructure monitoring, debugging and optimization (2008-2014)
  • Artemis; tools evolved from Artemis have been used by Hotmail to improve cluster operations.
  • Fay, a distributed cluster monitoring system.
• Large-scale applications (2008-2014)
  • training the Kinect body-parts classifier (Microsoft Research Technology Transfer Award in 2010)
  • branch-and-bound on a cluster
  • Sketch, a big-data spreadsheet.
• Programming languages and compiler architecture. I have contributed to several compilers: SQL to DBSP, DBSP, DDlog, SQL to DDlog, P4, CosmosDB SQL, DryadLINQ, the compiler forest, CASH (Compiler for Application-Specific Hardware), and the PipeRench compiler.
• Computer security (2005-2011):
  • Control-Flow Integrity (CFI). Our 2005 ACM Conference on Computer and Commnications Security (CCS) paper on CFI has received the 2015 ACM CCS Test of Time award. CFI is now supported by all major compilers and microprocessors.
  • Software sandboxing (XFI)
  • Safe dynamic execution tracing
• Hardware synthesis, harware optimization, reconfigurable computing, and computer architecture (1997-2007)
  • CASH: A Compiler for Application-Specific Hardware
  • asynchronous hardware design
  • PipeRench is a virtualized reconfigurable hardware chip. I wrote the PipeRench compiler. PipeRench was licensed by two start-ups (SiliconMobius, and Rapport, which commercialized it under the name Kilocore); both start-ups are now defunct. The PipeRench paper received the 2014 ACM SIGARCH/IEEE-CS TCCA Influential ISCA Paper Award, recognizing the ISCA paper 15 years earlier that has had the most impact on the field. The paper was also selected in the retrospective ISCA@50 25-year Retrospective 1996-2020

Some interviews I took: Kurt Akeley, Chuck Thacker (later winner of the 2009 ACM Turing award), and Leslie Lamport (later winner of the 2014 ACM Turing award), Brian Kernighan and Bruce Maggs.

Some of my talks:

Some of my papers