We're hiring in our ML team! 💻
Learn more
Two days ago I tweeted a link to a blog post I wrote stating that I'd left DeepMind to start a non-profit research lab, focused on reducing greenhouse gas emissions. I've been absolutely blown away by the response! I was expecting maybe a couple of re-tweets. At the time of writing, the tweet has been retweeted 470 times, liked 2.1 thousand times, and I've had over 200 messages! To be entirely honest, I'm slightly overwhelmed! (in a good way!)
I'd like to answer some of the great questions that people have asked over the last few days.
I'm at the very beginning of setting up this non-profit! I haven't settled on a name or precise company structure, I don't have a website or office, and don't yet have any funding secured!
I do have some friends with whom I've been discussion this non-profit for the last few months; some of these friends might get involved in the non-profit, although nothing is set-in-stone yet.
I've also been talking to the UK Met Office and National Grid for a little while, and - although nothing is certain yet - it looks very likely that my first project will be forecasting PV in collaboration with the UK Met Office and National Grid. I'm convinced that solar PV forecasting is likely to have significant climate impact, and it's a good fit for my skills and interests.
Not yet! But I will set one up ASAP. In the mean time, please keep an eye on my twitter feed :)
Peckham, London, UK :)
OK! Please see the Solar PV nowcasting blog post.
I don't have funding yet but I plan to apply to the Network Innovation Allowance soon. This will provide enough money for my time and some compute resources for a year or two; but the NIA unlikely to provide enough money to employ anyone else for the first year or so, I'm afraid - sorry!
The amazing response to my tweet made it pretty clear that there are loads (thousands?) of highly skilled folks out there who are really keen to help fix climate change. This is fantastic news!
For the solar PV nowcasting project, I'd love to talk to anyone who has an interest in PV, machine learning (especially video prediction), battery scheduling, grid optimisation, and weather forecasting.
I'll probably spend most of January 'just' talking to people :) (That's what you're supposed to do at the start of a startup, right?!?)
At least until the end of January, I doubt I'll get any significant code written. But, once I do start writing code, I'd love people to critique my preliminary results, and help bounce ideas around. I'll blog regularly with results.
In the medium term, we plan to put together a dataset and tool chain so anyonecan try their hand at PV nowcasting. This is really important to me: I want to enable everyone from teenage ML hot-shots to multi-national corporations to try to improve on the state-of-the-art for PV nowcasting. If they succeed in beating the state-of-the-art then I'd like to help get that algorithm into production so it can start making the world better ASAP :)
Unfortunately, I probably won't be able to provide meaningful projects for the majority of people who very kindly reached out to me. Which raises the question: How best to allow this awesome group of people to get stuck into helping build stuff which helps to fix climate change?
Would a database of climate mitigation projects be useful? If you're looking to work on a project (either paid or unpaid), then you'd look through this database to see if anything takes your fancy. If you're involved in a project, then you could ask for help. The emphasis would be on open-source projects which help to reduce emissions; but perhaps we'd also allow closed-source projects to advertise? Does that sound useful? If so, would a newsletter be best? Or a blog, which perhaps also publishes interviews etc. with people involved in climate change mitigation? Or a listing on github? Or a forum? Or all of the above? Or does this already exist?
For the first year or two, I'd like to focus mostly on solar PV nowcasting; and I'd like to deliberately keep the lab quite small, so that we can be experimental and agile without jeopardising lots of peoples' jobs!
If the solar PV nowcasting project goes well, and if the open-source approach has proved useful, then I'd like to scale up to a few other projects. This is when the lab would start raising significant amounts of money and employing people to work on a range of projects (all open source; all laser-focused on fixing climate change at scale).
I'd like to hold off on any press coverage at least for the next few weeks, please! I'd at least like to have settled on a name and set up a mailing list before any articles are published!
To be honest, I'd feel much more comfortable waiting until this lab has actually had real-world impact before doing any significant publicity. All I've done so far is write some blog posts and had a few meetings - stuff that the climate totally doesn't care about!
