TL;DR This job is for you if you enjoy:
- the incredibly exciting task of building the first foundational models for manufacturing
- working at the crossover of software and hardware (we are truly full stack and often are in factories)
- dealing with horrible messy real-world datasets and being high IQ to achieve SOTA performance whilst adhering to severe computational constraints
- delivering real impact on how we manufacture the world around us
About Matta
Matta is creating industrial AI for factories. We are a group of engineers, scientists, and company builders developing manufacturing foundation models (MFMs) and a general-purpose manufacturing OS to improve the sustainability of manufacturing today and to unlock the unimaginable technologies of tomorrow.
What we actually do:
- Build insane AI models that understand the physics of manufacturing processes (think futuristic stuff like machines that watch and correct themselves on the fly or even learn how to use brand-new, never-seen-before materials autonomously).
- Research new foundational models for manufacturing enabling users to install vision-based inspection and quality control AIs in less than 24 hours! (for context the industry standard is about 6 months).
- Think really hard about how manufacturing will look in 10+ years and build a brand-new intuitive OS and interfaces for orchestrating factories, monitoring machines, controlling production, etc. All with AI.
For context, we are a small dynamic team from Google X, Microsoft, Cambridge, MIT, Imperial, BBC R&D to name a few. We work a lot and are all close friends – come join the family!
AI Scientist @ Matta
We’re looking for an AI Scientist to join us in creating a brand-new class of foundation models designed specifically for manufacturing. You’ll get to tackle exciting challenges like:
- Inventing new capabilities, such as building models that understand causality – not just predicting what went wrong, but explaining why (e.g., “The temperature here was too high, mate. Lower it in this spot and you’re sorted!”).
- Boosting data efficiency and robustness to achieve top-notch accuracy with minimal data. Think teaching our models to learn from a handful of examples and transfer their smarts from one manufacturing setup to another like seasoned pros.
Oh, and here’s the kicker: all of this needs to work in real-time and run on low compute systems (because let’s face it, factories aren’t exactly decked out with H100 clusters). To give you an idea of the variety you’ll be diving into, one day you might be monitoring beer bottles, and the next, solving challenges with aerospace components. It’s a proper mix making life fun!
This isn’t your typical gig – expect to work hard and collaborate with an exceptional team of top engineers, scientists, and builders, all obsessed with embodied AI reshaping the manufacturing world.
About you
- A machine learning mage: You write clean, scalable, high-performance deep learning code. You take pride in your craft. Proficiency in PyTorch, Tensorflow, Jax, or equivalent is a must (ideally PyTorch).
- Knowledge on some of the latest good stuff from transformers to VLMs and NERFs.
- Happy to get down and dirty with cleaning our own in-house datasets (no nice benchmarks here) and building custom dataloaders, dataset classes, preprocessing functions etc.
- An understanding of classic signal processing approaches and computer vision techniques is super helpful (e.g. Fourier, convolutions, cross-correlations, Bhattacharya distance).
- You’ve led projects or research from start to finish, overcoming hurdles to create AI for products or high-impact publications (in startups? even better) plus you are a bit batty and want to learn more!
- People person: you love working in teams, sharing knowledge, mentoring/teaching others, and embracing feedback (and yes, you’re always game for a team Lego-building break).
- You’re adaptable and thrive in dynamic environments where every day brings new challenges.
Above all, we are looking to build a team of creative, outside-of-the-box thinking problem solvers. For this, candidates should be:
- Capable of setting and reaching ambitious goals, and taking a project from concept to completion.
- Self-directed, self-organised, proactive, and comfortable in a dynamic, multi-disciplinary work environment.
- Team players with excellent communication and interpersonal skills.
- Enthusiastic tea or coffee drinkers who enjoy a good biscuit or two and a good slice of cake (this is the most important criteria).
What will impress us
- Being proactive. Building or sending us example AI or data related projects – even better if related to the role.
- Whether you’re a chess grandmaster, a world champion at tiddlywinks, or a badass musician, we’d love to see the hobbies, skills, or activities you take to the next level – we’re always impressed by those who go all in!
- Clear passion in our mission, not a random reason to work with us.
Matta is an equal opportunity employer and encourages individuals from all backgrounds to apply. We offer competitive compensation, the friendliest of teams, an insanely stimulating work environment, and the opportunity to be a part of something massive.
Location
Our London office, 5 minutes from Old Street Station. Think vibrant startup attic vibes with a team that knows when to laugh and when to focus.
How to apply
To apply, please send us your CV and covering letter to careers@matta.ai or use Linkedin. Or if you are a little bit of a rebel (we like rebels) find some other way to apply.