Medior sEMG & Biosignal Machine Learning Scientist

About the role

We are looking for a medior scientist to develop methods that decode high‑density surface EMG (HD‑sEMG) into continuous force, joint kinematics and motor intent. The work combines experimental human data, wearable hardware and machine‑learning models to enable anticipatory control for our humanoid robots.

Key responsibilities

• Design and execute experiments to quantify the anticipatory information (30‑100 ms) in sEMG before movement.
• Develop signal‑processing pipelines for multi‑channel EMG, including filtering, artifact handling, normalization and channel‑quality assessment.
• Build and evaluate machine‑learning models that translate HD‑sEMG into force, torque and intent in real‑time or near‑real‑time settings.
• Investigate the impact of sensor placement, fatigue, session drift and other sources of variability on model performance.
• Collaborate with the robotics ML team to integrate decoded intent into the robot intelligence stack.

Required profile

• 3+ years of professional experience with EMG or other electrophysiological biosignals.
• Strong background in signal processing for noisy, non‑stationary physiological data.
• Experience developing machine‑learning models for time‑series or sequential data.
• Proficiency in Python and practical use of PyTorch or JAX.
• Hands‑on experience designing, running and analysing controlled human experiments.

Required skills

• Python programming
• PyTorch or JAX
• Signal‑processing techniques for EMG
• High‑density sEMG acquisition and preprocessing
• Machine‑learning for time‑series data
• Real‑time inference and latency optimisation

What we offer

• Unlimited paid time off and flexible working hours.
• Visa sponsorship, relocation assistance and on‑site work in Prague.
• Complimentary lunch, coffee and snacks daily.
• Top‑spec GPU workstation, cluster access and hands‑on time with robots.
• Conference budget and opportunity to publish open‑source contributions.