Machine Learning Software Engineer

THE ROLE

You would join our quant team, alongside our data scientists and ML engineers. The team owns several forecasting and optimisation systems that sit in the core of the product: site and portfolio-level forecasts of PV generation and load, calibration of our asset models against how assets actually behave, detection of what equipment sits behind a meter, and the optimisation that turns all of that into dispatch and trading decisions.

This is the engineering half of that work. You would own and improve the production systems those models run in, and the frameworks the team builds them with. The models themselves stay with the data scientists.

For applicants based in Vienna; traveling to Antwerp, Belgium once a month is expected. Applicants in Belgium will also be considered.

WHAT YOU'D WORK ON

Shared with the other ML engineers, and priorities move around, so read this as examples rather than a fixed list.

> A backtesting framework. Judging a candidate forecaster today means shadow- running it in production for weeks. Proper data and model versioning, historical replay, and evaluation in euros rather than error metrics alone would let us settle those questions in an afternoon.

> Engineering practices inside a data science team: testing, structure, code review, reproducibility, and whatever keeps research code alive once it reaches production.

> The MLOps lifecycle. Data and model versioning, retraining, backfills, and the Airflow jobs that carry them.

> Operational stability. A good share of the team's time currently goes into fixing what broke. Instrumentation, alerting and sane failure behaviour give that time back.

> Meteo data ingestion, storage layout and query performance.

> Scalability. Assets under management should grow sharply and the pipelines need to hold up without a rewrite each time.

> Research quality of life, including squeezing more out of our backtests and Ray.

WHAT WE'RE LOOKING FOR

> 2–5 years building and running production systems. Data engineering, backend or ML platform work all count. What matters is that you have operated something real, not only built it.

> Strong Python.

> Orchestration and cloud infrastructure. We run Airflow on GCP.

> Sound instincts on data modelling and storage, and a working understanding of database schemas and data structures.

> Enough ML familiarity to reason about training pipelines, feature computation and evaluation. You do not need to have trained models professionally. Initiative on engineering quality. Nobody here will hand you a specification for it.

> Energy domain knowledge is a plus, not a requirement.

A PhD is not required and we would rather see production engineering mileage.

Applied research backgrounds are welcome, but this is an engineering role and will be interviewed as one.

WORKS CLOSELY WITH

Data scientists and ML engineers on the quant team

STACK

Python, Airflow, GCP, Ray

WHY THIS IS WORTH DOING The bottleneck on the team is not ideas.

Time goes into fixing what broke and into proving that a change actually helped, and both of those are engineering problems. Solve them properly and the whole team gets its time back. You would also be shaping this function rather than inheriting someone else's version of it.

HOW HIRING WORKS

1. Send us your CV
2. Quick intro call
3. Coding interview
4. Technical interview
5. Culture check with the founders
6. Offer

Interested?

Send a CV and a short note on something you built and then had to keep running.

nikolaus@beebop.ai