As a founding hire, you will architect and train Adexis’s core visuo-tactile foundation model. You will own the entire pipeline, from the tokenization of visual and tactile signals to large-scale pre-training and downstream adaptation. This role defines how robots understand the physical world through touch, setting the scientific direction for physical general intelligence.
Member of Technical Staff – Pre-Training at Adexis
Machines can speak and see, but they are still numb. Adexis is changing that by building the visuo-tactile foundation model for Physical AI. As a founding Member of Technical Staff in London, you will architect the "somatosensory cortex" for robots from the ground up. Backed by leading VCs and researchers from Anthropic and DeepMind, this is a rare chance to own the training recipe for a model that truly understands contact physics. If you have trained large multimodal models and want to define a new field with 0.5-2.5% equity, this is your next move.
Want to apply for this role?
Jack finds you jobs at companies like Adexis. Talk to Jack to get considered for roles that fit what you're great at.
Location
London, United Kingdom
Compensation
£150k-£225k + Equity
Company
Adexis
Role overview
Founded in July 2026, Adexis is a London-based frontier robotics lab. We are solving dexterity by building the somatosensory cortex for Physical AI. Machines can speak our language and see our world, but they are still numb. Adexis will bring them the power of touch on the path to physical general intelligence. We are backed by leading VCs and angels from Anthropic, Google DeepMind, and ElevenLabs.
What you will do
- Design and own the end-to-end model architecture, determining how visual and tactile signals are tokenized and fused into a single learned representation.
- Develop a staged training pipeline involving large-scale pre-training on passive data, proprietary contact-rich data, and specialized downstream robotic adaptation.
- Establish the evaluation framework for tactile understanding and set the scaling roadmap as compute and data resources expand significantly over time.
Who this is a fit for
- Proven experience training large-scale multimodal or sequence models from scratch, with deep knowledge of data curricula, ablations, and debugging silent failures.
- Strong technical instincts for multimodal fusion and a track record of shipping models where non-visual modalities measurably improve system performance.
- High-ownership mindset with the ability to define architectures and make decisive technical choices when existing literature offers no clear direction.
Why this role is remarkable
- Join a founding team of outliers inspired by the Manhattan Project and Bell Labs to solve dexterous manipulation—the unsolved “missing piece” of Physical AI.
- Work alongside researchers from Anthropic and Google DeepMind with the backing of top-tier VCs to train massive, novel multimodal models from scratch.
- Enjoy massive scientific ownership and autonomy, defining a new field of visuo-tactile intelligence with a significant equity stake in a high-growth lab.
How Jack & Jill work together
Jack gets to know what you're great at and what you want next, then searches 15 million jobs daily and helps you discover roles at companies like this.
Meet Jack
What happens next?
Jack’s an AI agent for job searching and career coaching. He works for you.
Jill is the AI recruiter working for the company. She recruits from Jack’s network.
If your profile’s a match and Adexis wants to meet, Jill will make the intro. In the meantime, Jack will send you excellent alternatives.