Banseog note · 2026-10-26
A San Francisco company teaching robots dexterity from human muscle signals
Early stage — research-led, hiring core research staff one role at a time
- Relari describes its mission as robot dexterity learned from human biomechanics, combining video demonstrations with wearable surface electromyography (EMG) armbands to transfer force control, grip patterns and stiffness to robots, toward a foundation model for general-purpose manipulation.
- Its website lists Y Combinator, General Catalyst, Soma Capital and others as investors, and its job posting says the founders come from MIT, NVIDIA, Pony AI and Dexterity.
Decisions within a year
- Hiring as of October 26, 2026 the official careers page lists 1 physical-AI-related opening (Research Engineer, Human Biomechanics), full-time on site in San Francisco (Banseog collection).
People they will look for
- Research Engineer, Human Biomechanics — linking muscle activity to external forces is the core of the company's approach
- EMG, tactile and force signal processing (within the same posting) — the role calls for methods to process multimodal sensor data
- Biomechanics experiment design (within the same posting) — effects must be validated in experiments with robot-learning researchers
For hiring managers
In Korea, rehabilitation and wearable robot companies, robot hand and manipulation research groups, and university biomechanics and neuroengineering labs look for the same EMG, biosignal and motor control talent. The role is on site in San Francisco and requires relocation.
For job seekers
The role suits researchers who have collected and analyzed EMG or motion capture data themselves and handled signal processing and machine learning in Python. It requires relocation to the US.
An analysis written from public sources. Sources: Company website (relari.ai), official job posting (Research Engineer, Human Biomechanics), Banseog Physical AI official careers collection 2026-10-26 (1 opening).
Roles hiring now
Skills asked for most
Open jobs
Banseog checks official company careers pages every day. A job is marked closed only when the official page returns a removal (404 or 410).