ML Hydrologist, Flood Forecasting

Híbrido (4 días/semana) · Zurich (Suiza) · Mid · Jornada completa

¿Quieres más ofertas como esta?

Crea tu perfil gratis en PyHires: verás tu encaje con cada oferta y recibirás un email semanal con las que mejor encajan contigo.

Crear mi perfil

Resumen

Modalidad de trabajo
Híbrido (4 días/semana)
Ubicación
Zurich (Suiza)
Modalidad
Jornada completa
Seniority
Mid
Idiomas
Inglés
Abierta a candidatos de
Toda Europa (UE y fuera de la UE)
Empresa
Riverkin
Publicada
19 nov 2026

Sobre el puesto

Riverkin, a VC-backed ETH Zurich spin-off, makes global ML flood models accurate in the small, fast catchments where flash floods happen, using a sensor network it deploys itself. As ML Hydrologist you will own the hydrological modelling end to end, reporting to the CTO and working with the team's rainfall radar and nowcasting hydrologist. Your models will power early warning systems for hydropower plants and first responders. What you'll do - Adapt state-of-the-art hydrological ML models (LSTMs, Transformers) trained on large-sample datasets to customers' catchments - Use live streamflow data from the company's own sensors in previously ungauged catchments - Set the technical direction: decide when to adapt pre-trained models or train your own, from experimentation through validation to production - Integrate rainfall nowcasts into your flood models - Put models into production with software engineers and monitor their performance What you bring - Experience building and validating hydrological ML models across many catchments and taking them into real-world use - Validation across unseen catchments and time periods, including flood-peak magnitude, timing and predictive uncertainty - Practical deep learning for time series in PyTorch or a comparable framework - Strong Python, with maintainable code for data processing, modelling and reproducible experiments - Clear communication across hydrology, software, hardware and field operations Nice to have - Publications or open-source work on large-sample hydrology, transfer learning or ungauged basins - NeuralHydrology, ERA5 or similar global forcing datasets - xarray, Dask and Zarr for large hydrometeorological datasets What they offer - A permanent, full-time (100%) contract - Hybrid: four days a week at the Zurich office, one from home - GPU resources, and encouragement to publish and contribute to open tools such as NeuralHydrology and Caravan

Requisitos

Basada en:Python

Imprescindible

  • PyTorchEsencial

Deseable

  • Dask

Sobre Riverkin

Riverkin is a VC-backed ETH Zurich spin-off that builds flood forecasting for small, fast catchments where flash floods happen. It adapts hydrological ML models trained on thousands of catchments to local conditions using its own field-deployed sensor network, powering early warning systems for hydropower plants and first responders.

Stack

  • Python
  • Dask
  • PyTorch

Empresa