2026 UC Open: Day 1
Conference information/Schedule
Introductions
Carolyn Caizzi, UC Berkeley Library
- Library is the original open source repository for information.
- 20% about code, 80% about people, regulations, etc
Jarrod Millman, UC Berkeley
- Berkeley has a history of open source (FreeBSD)
- From Berkeley Institute for Data Science (BIDS)
- Interested in security
- Involved in scientific python (numpy, scipy, etc)
- Open source biggest challenges
- Funding
- Security
- Burnout
- Emphasis: community rather than code
Keynote Speaker: Open Software Entrepreneurship
David Charron, UC Berkeley Business School
- NSF looking at open source projects from academia
- Limited experience with running open source projects
- POSE Program
- (Example) FESTIM: Open source tool for fusion applications.
- Published open source rather than patent
- Blind to the people in the community
- NSF interested in increasing access to research funding by NSF
- People want to know how and have an implementation
- Scientists are open sourcing projects with minimum effort
- Put code online with no documentation or info who is using it
- Now PESOSE, Pathways to Enable Secure Open Source Ecosystems
- Added security
- Similar language to business, but with less “product” connotations
- (Example cont) FESTIM
- 100 people interviewed, get to know customers
- Learned why people choose their project
- 2 bus problem: What happens if the two founders disappear?
- Want to grow to many contributors
- No open source project is alone in the marketplace
- Natural Capital Alliance, Stanford
- Ecosystem map: Platform, funder, users, contributors
- There are always saboteurs who don’t want your project to succeed
- Companies using software to make money, but never received kickbacks
- Move from contributors to maintainers
- BDFL: Benevolent dictator for life
- Difficult to transition (funding, time, motivation)
- Ecosystem map: Platform, funder, users, contributors
- Value proposition is not “free”
- There are tangible benefits to open source
- I-corps syllabus has good reference
- Recommend 3 person team
- Team Lead
- Community Manager
- Industry Mentor
Takeways:
- NSF POSE gives necessary skills to manage and grow open source projects
- Need pathway to sustainability
- Community growth
- Develop for the “customers”
- Can really think of it as a business with different terminology
Panel
Experiences from the NSF POSE program
Stephanie Lieggi (UCSC):
- Looped project into a larger project
- Think of it as a successful application
- Merging with Apache Arrow
- Collaboration cleaned up code base
- Managing transitioned from student maintained to industry
Sanjit Seshia (UC Berkeley):
- Interested in verification
- SCENIC project
- Language where every variable is a random distribution
- Paper
- Holding 2 day bootcamps to get up to speed on the language
- Have working groups to connect developers to users, also workshops
Andrew Kahng (UCSD):
- Participated in the OpenROAD project
- Open source limited paperwork/overhead
- Cooperate partnership was beneficial in avoiding overhead
Keynote: The Role of Foundations in Advancing Open Collaboration and Innovation
Nithya Ruff, Chair of the Linux Foundation
- AI began at universities, but absent from current discussions
- Foundations gives universities opportunities for academia and industry
- Allow for a neutral place for code and community can live
- Examples for academia to foundation
- Spark
- RISC-V
- Contrary to geopolitics involving semiconductors
- Ray
- Ceph
- TUF
- CHAOSS
- Community health analytics open source software
- Urgency
- Federal Funding Pressure
- Geopolitical tensions
- Collaboration playbook
- Industry
- LXF Mentorship Program (mentorship.lfx.linuxfoundation.org)
- Linux Kernel Mentorship
- Submitting a patch to the linux kernel
- Google Summer of Code
- Academia
- Published paper and contribute the artifact
- Create a foundation project
- Possible funding opportunities from eBPF
- Industry
Takeaways:
- AI is a large focus of the Linux foundation
- Common for academic projects to transition to a foundation
Breakout: Open Source Tools for Scientific Research
Unmapped Cities: Scaling Pedestrian Infrastructure Mapping with Tile2Net
Maryam Hosseini, Assistant Professor, UC Berkeley
Assistant professor at city planning and development
- Looking at data from cities for accessibility
- Interested in pedestrian infrastructure
- Third of US pop non-drivers, 95% of funding goes to automotive
- Lack of data on pedestrian infrastructure
- Use aerial imagery for pedestrian infrastructure
Jupyter Book: Next-generation Tools for Creating Computational Narratives
Chris Holdgraf, 2i2c, Jupyter foundation
- Housed underneath linux foundaton
- Multi-stakeholder community
- Allow for rich markdown text to be generated into documentation pages
- has references, figures, mathtex
- MyST is the underlying parser
From Silos to Standards: Open Data Modeling with LinkML
Nomi Harris, Lawrence Berkeley National Lab
- Lot of scientific data, but difficult to use
- LinkML: Link Modeling Language
- Provide framework for standardizing data formats
Reflections on Building a High-Performance Microarchitectural Simulation Framework
Heiner Litz, UCSC
- 80% of Google’s energy spent on GP CPUs
- Focusing on increasing instructions per cycle
- Designed microarchitectural simulator
- Typically written in C/C++ for simulations
- Scarab
- 100000 USD per year!
- Testing involves running the arch on many different workloads
- Majority of code is written by AI
- Non-permissive license become absolute as code and be regenerated
Takeaways:
- Two of the tools have definitions as code, looks like the way forward
- LinkML has applications for representing data for Dirtviz
- Attendee has a good comments
- If a community exists around a platform, its unlikely people will switch
Panel: Making a CROSS play
- Chinstrap Community: Has course for taking open source project commercial
- Open source is like a car
- Individual parts are the source code
- Buying the car include warranty, support, onboarding
- Making money from open source
- Open Core
- RedHat is a classic example
- Open core: Core source is open source, setting enterprise features
- Traditional
- Create a business and open source parts
- Value is the combination
- Consultancy
- Not likely to get investors
- Offer professional services based on the project
- Open Core
- UC Berkeley has OP3 policy where participating projects return 1%