What should an E. coli eat for breakfast?
I cofounded a company called London Biocompute. Since my last post in August, we’ve been polishing our automated microbiology lab and supporting growth-assay experiments for ARIA-backed researchers at KCL. Bizarrely, equipment born in my bedroom has started contributing to actual research.
Now I want to expand access to our lab by making it available over the internet through a remote cloud lab API that lets anyone run experiments from anywhere. What better way to kick things off than with a challenge?
If you read some of my posts from last year like this one, you'll know I can't resist things like ARC AGI, Kaggle competitions, TinyGrad bounties, and hackathons. I think becoming obsessed with an exciting problem is the fastest way to learn a technical subject.
So I’ve designed a devious biological challenge that will test your ability to perform media optimisation - the process of optimising the diet and living conditions of a cell. Finding the right mix of sugars, temperatures, and nutrients that lets a culture grow as fast as possible.
This problem shows up across biology:
Cultivated meat: Media cost is one of the biggest barriers to making cell-cultured meat economically viable.
Scale-up: building robust, explanatory theories of how cells respond to external conditions that actually transfer across scales, from microplate to bioreactor.
Non-model organisms: There are countless organisms with useful properties that we simply don’t know how to culture efficiently yet.
What I’m personally most excited to see is whether total outsiders to biology can compete with wet lab scientists and academics. Can you find a clever optimisation strategy that beats someone with years of benchwork intuition? Could you even vibe-code your way to the top of the leaderboard?
Here’s the actual mechanics of the challenge:
You’re growing E. coli.
You get a budget of experiments and a list of sugars and trace elements to choose from, each with a known cost.
In each API call using our biocompute library, you specify a media recipe - which sugars, in what concentrations, and at what ratios.
Our robot runs your experiment on real cells in our cloud lab and returns the OD600 growth curve and images of the wells.
The goal: achieve the highest growth rate for the lowest media cost. It’s not enough to just blast the cells with glucose - you need to find recipes that are both effective and cheap.
If this sounds interesting, sign up here to get notified when the challenge goes live in the next couple of weeks!

Just wrote a post on ARC-AGI :) Indeed the pull towards these challenges is real.
https://blog.danielsosebee.com/p/the-progression-of-the-arc-agi-frontier
Great idea to make a challenge like this! Signed up though I'm not sure whether I'll have time to participate. And I know nothing about media optimization, lol.