Be a shiny needle: crafting standout startup job applications
How to signal that you're a driven person who solves important problems.
Archy de Berker
Blog and notes on climate, data, software, and work.
Latest posts
The UK is wasting a lot of wind power
The UK discarded 6% of its windpower last year, at a cost of almost £1bn. What’s going on, and how can we fix it?
Net Zero is a race, not a destination
The route we take to Net Zero determines how much CO2 we emit, and how much warmer we get
Blockchain solves none of the problems with carbon offsets
Blockchain solves none of the problems with carbon offsetting, but carbon dioxide removal solves many of them.
What do energy suppliers actually do, and why do they go bust?
We don’t think much about what suppliers do – except when they go really wrong. What service are they actually providing?
Why data scientists should work on climate change
There’s never been a better time for data scientists to work on climate
Minimal Viable Developer: acing your first month in software
Get fast at contributing code and flourish in your first software role
Sustainable Energy Without the Hot Air in 2021
How does Britain’s power profile in 2021 compare to the plans laid out in Sir David Mackay’s 2008 Sustainable Energy Without The Hot Air?
Adding interactive Plotly graphs to WordPress
Tips for adding interactive Plotly graphs to Wordpress and optimizing them for mobile
Should we install West facing solar panels?
The UK’s peak energy demand is at sunset, so should we be orienting new solar panels towards the West?
Dodging pitfalls when transitioning from academia to industry
Moving to industry can be a fantastic choice for some people, but getting your first job can be hard
How would Hayek solve climate change?
Hayek’s 1945 essay The Use of Knowledge in Society gives a lucid account of how carbon prices can simplify efforts to understand our environmental impact.
Things I didn’t understand about starting a company until I tried
No matter how many books you read, it’s a different story when you’ve got skin in the game. Some notes from starting Cholla, a remote team building company, in Summer 2020.
What does GPT-3 mean for AI?
The biggest AI news of 2020 was the success of OpenAI’s monstrous new language model, GPT-3. In this post I summarize why GPT-3 has caused such a splash, before highlighting some consequences for companies building things with AI.
Organizing applied machine learning research
Applied Machine Learning research is a bit different from standard software engineering. This post summarizes some of the things I’ve learned about leading applied ML teams.
Lockdown projects: Corona Calculator, Glass of Wine, and Weather Window
Lockdown in Montreal prompted a spree of building: The Corona Calculator, A Glass of Wine May Help, and Weather Window.
The customer is the enemy
It’s a bit of a cliché, but reading a tiny bit of military strategy can inform your thinking about startups.
Predicting the performance of deep learning models
It’s widely acknowledged that the recent successes of Deep Learning rest heavily upon the availability of huge amounts of data. Vision was the first domain in which the promise of DL was realised, probably because of the availability of large datasets such as ImageNet. The recent surge of simulators for RL further illustrates that as we \[…\]
Abuse detection on Twitter: a collaboration with Amnesty International
At the NeurIPS 2018 workshop on AI for Social Good we presented a piece of work we performed in collaboration with Amnesty International. We leveraged a mixture of crowdsourcing and deep learning to study the nature and quantity of abuse suffered by prominent women on twitter. At this point quite a lot has been written about the project: if \[…\]
How intelligent are our machines?
Human intelligence is remarkably correlated across domains. Do we mistakenly generalize this when assessing machines?
The Right Energy at the Right Time
During my ASI Fellowship I worked with Origami Energy, optimizing Demand Side Response programs for industrial energy usage.
Driverless cars and the attention economy
Driverless cars aren’t just a matter of convenience: they’re liberating hours of the day you could be buying things via Google!
Seeing the science in startups
If you squint a bit, building startups and doing science look kind of similar.
Building a Burns bot #2: Making a Slackbot
Here we take the Markov-chain based Burn’s Bot we created in the previous post and turn it into a living, breathing Slackbot to deliver Burns on demand.
Building a Burns bot #1: Markov models for poetry generation
How to generate synthetic poetry with the Markov models and a wee dram of whisky,
Tips from the other side: writing a thesis
After a relatively painless month or so of writing, I’ve finished my thesis. I’m yet to be examined, so whilst I can’t comment on how to write a good thesis, I’ve had a few thoughts on the process of getting from start to finish.
Three reasons you’re surprised that Trump won
Was Donald Trump’s election truly an upset? Some explanations from psychology for why we were more surprised than we ought to have been.
Deep Learning Practical 2: Decoding MNIST
MNIST (Mixed National Institute of Standards and Technology) is a database of handwritten digits. Compiled by Yann LeCun and colleagues, it’s a classic benchmark problem in machine learning. Pleasingly, you can write a model in Tensorflow that does a decent job of decoding the digits. Having got to grips with Python and some of the Tensorflow \[…\]
Deep Learning Book Chapter 3: Numerical Computation
This chapter was a bit of pot-pourri of things the authors wanted to tell us before we got on to machine learning proper. Much of it was straightforward, and then there were a few stingers, such as the bits about directional gradients. Typical problems we need to be aware of in using a digital computer to \[…\]
How others’ fortunes affect your happiness predicts generosity
This week we published a paper building on Robb Rutledge’s 2014 happiness model (my summary here), but asking whether we could also capture fluctuations in happiness that relate to what’s happening to other people around you.
Deep Learning Practical 1: Intro to Tensorflow
This week in the Deep Learning Reading Group we worked through some introductory exercises on Tensorflow, based upon the provided tutorials and edited by Zeb and I. It’s targeted at those with minimal Python experience, but who have programmed in Matlab. It’s largely self-explanatory; a collection of code examples with some discursive text, and a few snippets \[…\]
Deep Learning Textbook Chapter 2: Probability
As part of the Max Planck Centre’s Deep Learning reading group, I’m working my way through the soon-to-be released Deep Learning textbook. These posts are my informal notes for each chapter. Find my notes from chapter 1 here. Thanks again to Zeb & Toby for edits. This week we were back on slightly firmer ground, dealing \[…\]
Deep Learning Textbook Chapter 1: Linear Algebra
The Dolan lab group (and friends) is working through the soon-to-be-released Deep Learning textbook, authored by Goodfellow Bengio, and Courville, and somewhat miraculously available in html form prior to publication by MIT press. At a risk of exposing the depths of my ignorance to the internet, I’m going to try and publish my notes on each chapter \[…\]
Setting up Tensorflow for Mac OS X
Alternative method from the Asimov institute. This worked fine for me using an Anaconda install of Python on Mac OS X, except I had to remove the ‘–update’ flag when doing the installation with pip. Their version involves direct installation on OS X rather than installation on a virtual Linux machine via Docker, which is \[…\]
The Happiness Project
The Happiness Project is a project I’ve been involved in for just over two years, which began as a low-key attempt to explore why a UNICEF report found British children to be very unhappy relative to children in similar countries and blew up into a show involving 12 young adults and 6 academics performed at the Edinburgh festival.
What’s the deal with modafinil?
Scientists need to stop hyping smart drugs. We don’t know if or how they work and what the long term consequences of use might be.
Running deep reinforcement learning models
Deep Mind’s Nature paper has created an understandable buzz. In it they create an algorithm than can learn to play a huge slew of Atari computer games better than a human expert. Particularly compelling are the videos, which I encourage you to check out. One delightful feature of the paper is that they provide the \[…\]
Substantia Nigra and Reinforcement Learning
Robb Rutledge and I have a piece just out in the Journal of Neuroscience. It’s a Journal Club article – a ‘short, critical, review’ of a recent paper published by Ramayya et al in Michael Kahana’s group at University of Pennsylvania. The paper concerns a simple learning experiment performed in a group of people who \[…\]
How can we build computational models of happiness?
My colleagues recently published a seminal paper examining the dynamics of happiness within a prediction-error framework (Eh? What does this mean? Read on to find out).
Einstein in Berlin
At the end of July I enjoyed a few days in Berlin, attending the 4th (and, sadly, final) Einstein Fellowship Symposium on Decision-making, organised by my supervisor Ray Dolan. After 2 days of exceptionally high-quality talks and posters, we finished with a superb boat trip around Potsdam, a rather plush suburb riddled with lakes and \[…\]
SfN 2013- San Diego
This November, me and a gang from Queen Square descended, along with 30’000 other neuroscientists, upon San Diego, California. Amongst this maelstrom of intimidatingly brilliant minds I was to be found presenting a poster on happiness in a social context, which you can read here if you’re interested.