Senior Machine Learning Engineer (m/f/d)

Personio; Mainz; Hybrid; EUR 70,000 - 100,000; Full time

Benefits offered by this job

Monthly voucher for shops or Urban Gym

Office with height-adjustable desks

Weekly team fitness sessions

Yearly company offsites

Job summary

Personio is building a health-tech feedback system that evaluates movement quality directly from keypoints, with edge deployment on mid-range Android or iOS devices. You will own end-to-end development of the feedback model, from data collection and annotation strategy to training, evaluation, and deployment, while collaborating with physiotherapists and doctors to define clinical benchmarks.

You will optimize body and hand tracking models, research new approaches, and implement scalable

Qualifications

• Proven experience building, optimising, and deploying Deep Learning/Computer Vision models to production with end-to-end ownership.
• Experience with time-based signals for action quality/error assessment from video.
• Practical experience applying transfer or few-shot learning to generalise model predictions to unseen motion or fault classes.
• Experience with model compression, quantisation and pruning for edge/resource-constrained deployment.
• Strong Python engineering skills with PyTorch or TensorFlow and clean code practices.

Responsibilities

• Identify, extract, and process data from relevant sources, and shape the annotation strategy.
• Research new approaches and turn them into solutions.
• Design and implement deep learning pipelines with a focus on computer vision from data through training to deployment.

Skills

Deep Learning

Computer Vision

Edge deployment

Transfer learning

Model compression

Python

Tools

PyTorch

TensorFlow

Job description

Your mission

Our health apps guide patients through rehabilitation exercises, watch them train via camera and tell them what to correct. Today that runs on biomechanical rules written by hand for each exercise. Your mission: Replacing them with a model that evaluates movement quality directly from keypoints, handling the following challenges:

• Scale: Hundreds of exercises, several fault classes each, and what counts as a fault shift with the impairment. Adding new exercises or error classes should stay cost efficient – with one model per exercise we don’t get there, so generalization is needed
• Constraints: Edge deployment directly on mid-range Android or iOS devices. You will handle noisy keypoints, occlusion, dropped frames, and low-latency edge processing to give patients instant feedback to their exercise repetition.
• Reference: Correct execution is a clinical judgement, and our physiotherapists and doctors are here to define it with you. Turning that judgement into something a model can learn from, and into an evaluation we can trust, is part of the research question

Having end to end ownership for the feedback model also includes:

• Identify, extract, and process data from relevant sources, and shape the annotation strategy
• Research new approaches and turn them into solutions
• Design and implement deep learning pipelines with a focus on computer vision, from data through training and evaluation to deployment.
• Create tests and evaluations that measure model, but also production performance

With a data science team of three, you will also do hands‑on work across the team, especially supporting the optimisation of our body and hand tracking models.

Your profile

• Proven experience building, optimising, and deploying Deep Learning/Computer Vision models to production with end‑to‑end ownership.
• Experience with building models on time‑based signals for action quality/error assessment for closely related fields like assembly‑step monitoring, surgical skill assessment or ergonomics scoring from video.
• Practical experience applying transfer or few‑shot learning to generalise model predictions to unseen motion or fault classes.
• Experience with model compression, quantisation and pruning for deploying models on edge or resource‑constrained devices.
• Strong Python engineering skills with deep expertise in PyTorch (or TensorFlow) alongside clean code practices, modular architecture, and testing principles.

Strong candidates may also have

• An eye for how human bodies move: background in biomechanics or physiotherapy
• Experience training pose estimation or hand tracking models from scratch
• Worked in a medical or otherwise regulated setting

Why us?

• One Team One Dream: Remember being the person who did all the work in university group projects? Imagine a team made up entirely of those people - working together with mutual support and zero ego games.
• Freedom & Responsibility: You know your calendar and your deadlines. Feel free to take a Yoga class on Wednesday morning or go the dentist on Monday afternoon - and finish your tasks on Saturday to make it all happen in time. Take responsibility and gain freedom to accomplish things your way.
• Best Ideas Win: We make decisions based on facts and strong arguments, never hierarchy or titles. You are not just allowed to speak up - you are expected to.
• Get Things Done: An idea is worth nothing until it survives contact with a real user. So we ship early, fail fast and learn from it. That is how our small team gets things done in weeks that larger organisations take years to finish.
• Level Up: You own your growth. We support you with direct feedback, courses, mentoring, and the opportunity to publish the research you work on.

Benefits:

• Monthly voucher for shops, groceries or an Urban Sports Club membership
• Office with height‑adjustable desks, free parking, air conditioning, cool drinks and good coffee
• Weekly team fitness sessions (e.g. Pilates, Tabata), darts and table tennis
• Monthly fun events and yearly company offsites