Senior Data Scientist

Location

Switzerland: Solothurn, Hybrid

Start

By arrangement, permanent

About us

QUMEA is a pioneer in digital mobility monitoring for healthcare institutions. We stand for innovation and progress in healthcare and set new standards in patient safety. Our radar-based system measures human movement in patient rooms contactlessly and anonymously. Care teams automatically know around the clock how their patients are doing and are informed in real time when help is needed. This digital mobility monitoring increases patient safety, supports care teams in their daily work, and delivers relevant clinical insights.

What drives us? Supporting healthcare professionals in proactively managing risks and improving outcomes— with a solution that works safely, efficiently, and discreetly. To achieve this, we develop not only products, but also knowledge and technologies that are ahead of their time. To strengthen our Algorithmics team, we are looking for you as a Senior Data Scientist & Software Engineer to further develop our data pipeline.

Tasks

Your mission

As a Senior Data Science & Software Engineer, you will develop new, clinically relevant capabilities for patient monitoring from our radar data. You will evaluate creative new approaches and translate them into robust, measurable models that make a real impact in everyday care - for example, lone/presence detection, gait analysis (movement patterns), and new features such as delirium.

Your focus is scientifically curious while being hands‑on and product‑oriented: you don’t just build prototypes, you integrate solutions cleanly into the product landscape with sound architecture, clean code, and an engineering mindset (Java & Python).

Your responsibilities in detail

• Development and validation of models/algorithms for use cases such as:
  • Gait analysis and movement‑pattern recognition (gait characteristics, stability, deviations, trend analyses)
  • New features such as delirium, complex risk indicators, clinical "events"
• Experiment design & measurability: define metrics, offline evaluation, golden sets, reproducibility, performance/robustness.
• Feature engineering & representation learning: derive meaningful representations from radar data (including domain understanding).
• Production‑ready implementation:
  • Transfer prototypes into clean, maintainable software components
  • Clean Architecture, testable code, code reviews, refactoring.
• Integration into our existing pipeline/services (Java/Python)
• Evaluate new approaches for radar‑based patient monitoring (classical ML, deep learning, probabilistic models, first‑principles and hybrid methods).

Requirements

What you bring

• Several years of experience as a Senior Data Scientist / ML Engineer / Software Engineer (or comparable), with proven production ML systems.
• Strong ML knowledge (supervised/unsupervised, sequences/time series, anomaly detection, classification/regression) and solid understanding of statistics and evaluation.
• First‑principles thinking: you don’t just "apply" models—you derive them, challenge assumptions, and combine them with domain knowledge (hybrid approaches).
• Very strong programming experience and an engineering mindset:
  • Python (analysis, prototyping, modelling)
  • Java (product integration, clean services/components)
  • Testing, code quality, maintainable design, performance awareness.
• Experience across the full model lifecycle: understand data à build a model à evaluate à bring to production à
• Scientific curiosity combined with pragmatism: form hypotheses, test them experimentally, and deliver results.

Nice to have

• Radar/sensor experience (radar preferred; alternatively similar modalities with demanding signal/time‑series characteristics).
• MLOps / ML‑Pipelines: training/deployment automation, model registry, monitoring/drift, reproducible experiments.
• Experience with scalable data/feature pipelines, streaming/batch processing, observability.
• Experience with robust benchmarking (regression suites, golden data replays, A/B tests).
• Domain know‑ho