OPEN SOURCE, AND WHOSE JOB IT TAKES

Open Source Leadership Summit

6–8 March 2018 · Fairmont Sonoma Mission Inn, Sonoma Valley, California · LAB

A panel at the Linux Foundation's Open Source Leadership Summit, on what open source does for deep learning and on the argument that intelligence in code puts developers out of work.

The Linux Foundation runs its Open Source Leadership Summit for the people who set direction rather than the people who file the pull requests. The 2018 edition ran from 6 to 8 March at the Fairmont Sonoma Mission Inn, an hour north of San Francisco, in a marquee pitched on the lawn with chandeliers strung along its spine. Jim Zemlin, Mark Russinovich of Microsoft, Erica Brescia of GitHub and Marten Mickos of HackerOne were among the speakers that week.

deepkapha.ai was invited to join a panel discussion alongside professionals, entrepreneurs and investors from Google, Amazon, Microsoft, LinkedIn, VMware and IBM. The subject was how open source technologies advance deep learning and AI, which in early 2018 was less obvious than it sounds now. TensorFlow was two years old. PyTorch was one.

The panel also took on the anxiety in the room, that pushing intelligence into code automates away the developer's job. Our answer was that disease, hunger and poverty still need solving, and the work is urgent enough that there is no shortage of it. The rest of the visit went to the lab's philanthropy: what it teaches, who it teaches, and why it does that in the open.

The unconference board settles a detail the professional photographs leave out. Tarry's session was written up in marker on the Linux Foundation's sign-up whiteboard: Deep Learning/AI Career with Open Source, Thursday at 2:30, forty-five minutes, run as a whiteboarding session rather than a talk. Under the title, in red, sat the question the room had come for. How do I fasttrack my learning path to AI? He signed it @deepkapha.ai.

Unconference slots were not on the printed programme. The daily schedule screen listed the Vineyard Room as Open Spaces Unconference Session in every band from 11:40 to 5:00, with one instruction beside each: see registration desk to book. Speakers claimed their time by writing on the board. Tarry photographed it at 11:23 on Wednesday 7 March, with his own line already down for the following afternoon and a blockchain session booked into the 3:30 slot after his.

The venue map banner at the hotel entrance divides the Inn into working parts. Keynotes in the Pavilion. Breakfast, lunch and breaks under the Breezeway Lawn Tent. Breakout sessions across the Pavilion, Kenwood 1, Kenwood 2 and the Sonoma Valley Room. Unconference sessions in the Vineyard Room. The Oak Room kept as speaker and media lounge, the Barrel Room and Carneros Suite bookable at the desk.

Wednesday's keynote grid put the subject on the main stage hours before Tarry got to his own slot. Mark Russinovich, chief technology officer of Microsoft Azure, opened at 9:00 with a talk titled AI Owes Its Rebirth to Open Source and the Cloud. Edward Kearns, chief data officer at NOAA, followed at 10:25 on open data. The room that filled for both was a marquee pitched on the lawn, chandeliers strung from the canvas, screens down both long walls.

Buses left the Inn at 6:15 that evening for the summit reception. Tarry's frames from 19:00 onward put him in a winery barrel room off Arnold Drive south of Sonoma, badge still on, in conversation with delegates including Rao Mikkilineni of C3DNA. He had flown in two days early, through Frankfurt on the morning of 5 March and into the Sonoma grounds by four that afternoon.

The old site recorded this as San Francisco on 3 March 2018, and the source of that error is traceable: the event photographer's filenames are prefixed 20180303. The Linux Foundation's own page gives 6 to 8 March at the Fairmont Sonoma Mission Inn, an hour north of the city.

FROM THE FIELD · 24 FRAMES

Tarry Singh mid-sentence on the panel, hand raised to make a point, wearing the #machinelearning T-shirt.

The panel from the side of the stage, all five seated in a row against the summit backdrop, Tarry Singh second from the right in the #machinelearning T-shirt.

A session in the summit marquee, a speaker at the lectern beside a slide asking how to make the open source feedback loop spin faster, the audience seated in the dark below a ring chandelier.

The marquee from the back during a session, ring chandeliers down the length of the ceiling and the audience seated in rows towards the lit stage.

The marquee from the side, every seat taken, daylight through the windows along the far wall and screens either side of the stage.

A speaker at the lectern working through a slide of adoption percentages, the audience in the dark below the stage.

Photographs by Stan Olszewski / SOSKIphoto for The Linux Foundation, from the summit's own album, used under CC BY 2.0. The first three frames are the panel; the rest are the summit around it, and frames 4 and 7 are other people's sessions.

ALSO FROM THIS WEEK · 17 MORE

01 / 17

A framed sign standing on the inn's grounds, headed Private Event, carrying the Open Source Leadership Summit mark over a vineyard photograph and the line "Gather here nightly for drinks, snacks and good conversation!"
The Fairmont Sonoma Mission Inn at dusk, the low pink building behind lawn and planting.
The grounds between sessions, delegates sitting on the lawn under the trees in front of the inn.
Delegates seated through the Pavilion marquee at the Open Source Leadership Summit, a seven-person panel on the lit stage under the Open Source logo, chandeliers hung from the canvas ceiling
The Linux Foundation Unconference Sessions Sign-up whiteboard, Wednesday and Thursday columns filled in by hand, Tarry Singh's deep learning session written into the Thursday 2:30 slot
A full room of delegates in the summit marquee, rows of grey chairs, seats at the front marked reserved for speakers
The white marquee entrance to the summit Pavilion, Keynotes signed above the doors, sponsor banner beside it and delegates walking in
Close view of the sign-up whiteboard: Deep Learning/AI Career with Open Source by Tarry Singh, @deepkapha.ai, 45 minute session, how do I fasttrack my learning path to AI, whiteboarding session
A speaker at the Open Source Leadership Summit lectern in a breakout room, slide behind reading Collaboration Examples with TensorFlow, PyTorch, ONNX, Gluon, Apache MXNet, Kubernetes and Docker listed
Linux Foundation venue map banner at the hotel entrance listing keynotes in the Pavilion, unconference sessions in the Vineyard Room and the Oak Room as speaker lounge
The Wednesday schedule screen showing keynotes from 9:00 and five parallel session rooms, with the Vineyard Room column reading Open Spaces Unconference Session, see registration desk to book
The summit keynote stage, Open Source Leadership Summit backdrop in blue and white, bar stools set for a panel and a speaker at the lectern
Tarry Singh with a fellow delegate in a winery barrel room at the summit evening reception, conference lanyards on, oak barrels racked along the wall behind
The Sonoma Mission Inn grounds from above, white event marquees pitched on the lawn beside the fountain garden with the Sonoma hills behind
Delegates in the Inn lobby seen through the plastered arches of the colonnade
The summit lawn at dusk, the event tent lit from within, benches and lawn games set out on the grass
The Fairmont Sonoma Mission Inn courtyard before dawn, arcade windows lit along the ground floor and the bell tower above