Member of Technical Staff – Research
Read straight from Adexis's careers page, verified 1d ago.Seekless can tailor your résumé to it.
Seekless can write your résumé for this posting first.
Before you apply
Salary£150k – £225k
LocationLondon
Work modeon-site
Typefull-time
SeniorityLead
DepartmentResearch
EquityIncluded
Company size3+ people
PostedSep 30, 2026 · 1mo ago
Verified live1d ago
At a glanceSummarised by Seekless from the posting.
Trained large multimodal or sequence models from scratch and owned the recipe — the data mix, curricula, ablations, debugging silent failures
Have strong instincts for multimodal fusion
Trained manipulation policies and put them on real hardware — you know the specific pain of a policy that demos well and dies on contact
Know imitation learning deeply and know when reinforcement is worth the instability
Have a point of view on learning dexterous, contact-rich skills rather than pick-and-place
Teleoperation systems, learning from human demonstration, real-world RL
Benefits
Health insurance
Equity
About Adexis
Adexis is a London-based frontier research lab solving dexterous manipulation. Our mission is to bring machines the power of touch on the path to physical general intelligence, bringing about a future of abundance.
We are building a lean, talent dense organisation, inspired by history’s great groups — the Manhattan Project, Apollo, Bell Labs, Skunkworks — where aggregating outlier individuals drove collective output that changed history.
We are backed by leading investors, and researchers from Anthropic, ElevenLabs, Google Deepmind and more.
Member of Technical Staff — Research
Responsibilities
As a founding research hire, you could work across pre or post-training for our visuo-tactile foundation model. This might include:
•
Own the model architecture end to end: how visual and tactile signals are tokenized, fused or kept separate, and learned jointly.
•
Make the hard representational calls: what a contact signal is at the token level, how it aligns in time with vision, and how the two modalities trade off when they disagree.
•
Own the post-training stack: how we adapt a pretrained model into a control policy using teleoperated demonstration data.
•
Design the recipe: demonstration collection strategy, imitation and any reinforcement or self-improvement stages, and the full sequence from foundation model to deployable policy.
•
Close the sim-to-real and human-to-robot gaps: make a policy learned from human demonstration survive contact with the real world.
•
Define what good behaviour is and build the evaluation that catches a policy that looks good offline and fails on hardware.
You might be a fit if you
•
Have trained large multimodal or sequence models from scratch and owned the recipe — the data mix, curricula, ablations, debugging silent failures.
•
Have strong instincts for multimodal fusion.
•
Have trained manipulation policies and put them on real hardware — you know the specific pain of a policy that demos well and dies on contact.
•
Know imitation learning deeply and know when reinforcement is worth the instability.
•
Have a point of view on learning dexterous, contact-rich skills rather than pick-and-place.
•
Bonus: teleoperation systems, learning from human demonstration, real-world RL.
We seek a high ownership individual up to the task of defining this vision. This is a founding research role and will be compensated accordingly with equity.
Compensation & benefits
•
Salary: £150–225k
•
Equity: 0.5-2.5%
•
Private healthcare with AXA
•
Top of the line equipment
More roles at Adexis