Andrew Gaul
Systems programmer interested in entrepreneurship and technical leadership
opportunities in San Francisco and Tokyo.
Background includes work in distributed computing, machine learning, storage,
and free software.
Work experience
- Developed a machine learning platform for long-lived and repeatable experiments across many GPUs and systems
- Led implementation of fault tolerance, multi-file models, and management of many experiments
- Contributed to model cost estimation, system dump, web UI, and championed upstream contributions
| Consultant and Sabbatical |
|
2016 - 2017 |
- Consulted for clients on Apache jclouds, goofys, s3fs, and S3Proxy and
visited 17 countries across Asia and Europe
- Created a hybrid object store for managing data spanning
different clouds and geographies
- Implemented protocol translator comprised of an S3 frontend and an enhanced
Apache jclouds backend
- Led outbound efforts including pitching to investors, sourcing customer
prospects, and presenting at conferences
| Maginatics (acquired by EMC) |
Staff Engineer |
2013 - 2014 |
| Senior Engineer |
2011 - 2013 |
- Founding engineer for
MagFS
1.0-2.6, a cloud-based network file system which evolved into
EMC
CloudBoost
- Lead developer of object data storage, SQL metadata storage,
deduplication, garbage collection, backups, and snapshots
- Worked with the Apache jclouds community to improve
performance and compatibility across 15 different object stores
- Addressed database concurrency, management, and space-efficiency issues
and contributed to upstream MySQL projects
- Recruited and mentored engineers, supervised contractors, and interacted
with customers, investors, and partners
Storage engineer for
SteelFusion
1.0, a virtualized, WAN-optimized iSCSI storage appliance
- Contributed to core indexing structures, transactional persistence and recovery
of data, and overall system performance
- Established internal dogfood setup, deployed initial
customer installs, and evangelized failure-injection testing
File system engineer for
Atlas 1.0,
a deduplicating storage appliance which evolved into
NetApp
AltaVault
- Implemented algorithms for deduplication and data
structures for efficient indexing and storage of compressed data
- Improved data segmentation, snapshot, and garbage collection
algorithms, increasing performance and space efficiency
- Provided 2-10x speedups
via algorithmic improvements, assembly-level optimizations, and
implementation tuning
| Tabernus (acquired by Blancco) |
Software Engineer |
2004 - 2006 |
Free software
- Apache jclouds
committer and
PMC member (2013 - present),
Vice President (2017),
and GSoC mentor (2014, 2017)
- Implemented the B2 API, overhauled filesystem implementation, enhanced multi-part upload, and expanded the portable abstraction across seven object stores
- Reviewed more than 100 pull requests by over 40 authors and individually resolved hundreds of issues
- s3fs committer
(2019 - present) - accelerated readdir, rename, and write performance by 2-5x,
addressed concurrency and data loss issues, and improved POSIX compatibility,
S3 interoperability, testing, and documentation
- Author of S3Proxy, a virtual
object store allowing access to various storage backends via programmable
middlewares
- Project has more than 25 contributors and over 500,000 Docker pulls
- Author of
Chaos HTTP
Proxy,
Java httpbin,
JDBC lint,
Modernizer Maven Plugin,
and others
- Contributed to
compy,
goofys,
H2,
innodb_ruby,
mbpfan,
OpenSSH,
Percona InnoDB tools,
s3-tests,
Thrift,
and more
Education