Algorithm Developer (m/w/x)

RIZM; München; Híbrido; EUR 90.000 - 140.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

Equity

Hybrid work model

Team offsites/workation

Urban Sports Club

Descripción de la vacante

RIZM is seeking an experienced optimization engineer to own end-to-end energy planning models. You will work on capacity expansion, quarter-hourly dispatch, and production scheduling with MIP formulations, integrating regulatory rules and complex physics into scalable workflows.

The role emphasizes production deployment, AI-native tooling, long time-series data, and close collaboration with Product Engineers. Hybrid work from Munich is offered.

Formación

• Degree in a STEM field relevant to optimization.
• Deep familiarity with LP and MILP and related formulations.
• Production deployment experience with models used in production.
• Experience with JuMP and Gurobi/HiGHS.
• Experience with long time series data and time-series conventions.
• AI-native tooling experience (Claude Code, agents) is a plus.
• Ability to own complex optimization projects end-to-end.

Responsabilidades

• Own capacity expansion planning over multi-year horizons and quarter-hourly dispatch.
• Formulate physics for storage, thermal networks, transmission and contractions.
• Tighten big-Ms, explore decomposition, and ensure scalable models.
• Reduce data via time-series clustering and segmentation without losing accuracy.
• Incorporate regulatory charges, tariffs and energy market rules into constraints.
• Validate with real-world tests and provide diagnostics for infeasibilities.
• Interface with Product Engineers; run optimization jobs on Kubernetes; optimize time, memory, cost.
• Explain model schedules to customers when questioned.

Conocimientos

LP & MILP optimization

Production deployment experience

Julia or Python/C++ programming

JuMP/Gurobi/HiGHS familiarity

Data handling of long time series

Solver tuning & reformulation

AI integration: Claude Code & agents

Energy systems context knowledge

Educación

STEM degree (math, OR, CS, physics, energy systems, or similar)

Herramientas

JuMP

Gurobi

HiGHS

Python

Julia

TypeScript

Vue

Node.js

PostgreSQL

Kubernetes

GitLab CI

Descripción del empleo

Invest in Values. Produce in Freedom. Boost Europe.

Europe's plants run on gas, power, heat and steam – and all of them are being rebuilt at the same time. The hard part isn't ambition, it's arithmetic: which asset to buy, when to run it, what to hedge. Europe's largest companies (~2% of total energy demand) answer these questions with our Energy OS: a digital twin of the site, an optimizer that plans it, and agents that operate it.

The optimizer is yours. Twenty-year investment decisions, quarter-hourly dispatch, real tariffs, real market rules. You report directly to the CTO – and nobody stands between your model and the plant it runs on.

Tasks

Optimization Models End-to-End: Own capacity expansion planning over multi-year horizons, dispatch at quarter-hour resolution, and MIP production scheduling with sequence-dependent setup times.

Model Physics & Contracts: Storage, thermal networks, transmission, flexibility, self-consumption, minimum part loads, start-up costs – formulate what reality actually demands.

Make It Solve: Tighten big-Ms, scale models cleanly, decide where decomposition earns its complexity – and where it doesn't.

Smart Data Reduction: Time series clustering, extreme periods, segmentation and aggregation – without accuracy quietly disappearing.

Regulation into Constraints: Model grid fees, levies, tariffs, balancing power and intraday trading per country, and understand what they do to the objective function.

Prove It's Right: Test cases on real customer systems, fixtures that catch regressions, and infeasibility output an engineer can actually diagnose.

Interface & Production: Define inputs and outputs with the Product Engineers; own the runs on Kubernetes – solve time, memory and cost are your numbers.

Explain to the Customer: When a customer questions a schedule, you're the one who can say why the model chose it.

Requirements

Education & Background: Degree in mathematics, operations research, computer science, physics, energy systems engineering, business mathematics or a comparable STEM qualification.

Optimization: Deep familiarity with LP and MILP and the craft around it – formulation strength, relaxations, warm starts, solver behavior.

Practice over Papers: You've taken a model from formulation to production that others depended on.

Code & Solvers: Strong Julia – or strong Python/C++ and a real appetite to learn Julia (JuMP) fast. Hands-on with Gurobi or HiGHS, including reading logs and knowing when to reformulate instead of tuning.

Data Sense & Engineering: Comfortable with long time series, unit and sign conventions; tests, version control, code review and CI are second nature.

AI-Native: Claude Code, agents and subagents are part of your daily loop – a requirement, not a plus.

High Ownership, Low Ceremony: You take on the unclear. Energy domain knowledge helps but isn't required.

Nice to have: Stochastic or robust optimization, Benders or column generation, unit commitment, energy market modeling.

Share in the Success: Competitive base salary plus equity – we set your level during the process, not based on your CV.

Top-Tier Venture Capital Network: Direct access to the Point 9 and Earlybird networks – including formats for sparring, mentoring and further education.

Unforgettable Team Offsites (Workation): Once a year, the whole team goes on a workation. Including exclusive exchanges with our Visionary Voices (e.g. Robert Habeck, Kitty Mayo, James Vincent).

Corporate Health: Full flexibility for your fitness and health with Urban Sports Club.

Out to the Plant: Seeing the boiler you modeled changes how you build the product.

Our stack: Julia with JuMP on Gurobi and HiGHS, forecasting in Python, the product in TypeScript (Vue, Node, PostgreSQL), Google Cloud with Kubernetes for optimization jobs, GitLab CI, agents on LangGraph.

Small team, no layers between you and the decision. A model change that lands on Tuesday is in front of the customer on Thursday. In-person onboarding, then hybrid from Munich, Düsseldorf or Münster.

Not a research position without deployment – and not a place where solve time is someone else's problem.