Founding Applied ML / Quant Forecasting Engineer

Elcast.ai builds forecasting, ranking, and decision systems for European power-market trading, with a particular focus on identifying attractive opportunities in JAO month-ahead transmission rights auctions.

These auctions are economically valuable because transmission rights can gain or lose value based on future cross-zonal price differences, congestion dynamics, outages, network conditions, weather, generation, demand, and broader market structure. Our commercial challenge is to identify auctions that are undervalued relative to their future realized value before bid-gate close.

Over roughly three years, we have built a highly refined data, forecasting, and research platform covering ENTSO-E, JAO, third-party forecasts, Trading Economics, FX, holidays, and Copernicus medium-term weather data. In internal backtesting, our models have produced average monthly profitability as high as approximately 70% when trading volume is restricted to the highest-ranked JAO month-ahead auctions only. We care deeply about leakage control, architecture quality, and production discipline because the opportunity is commercially meaningful.

Why this role is different

This is not a generic startup ML role, a notebook-only data science role, or a pure MLOps role. You will join a substantial forecasting and decision platform with a PostgreSQL-backed data layer, structured feature and prediction pipelines, MLflow experiment tracking, batch-level provenance, and our internal web control interface.

The role is not to start from a blank page. It is to use, validate, improve, and extend an existing system where prediction batches, feature sets, model versions, walk-forward runs, validation status, and auction-level performance can be inspected visually rather than hidden inside disconnected notebooks or scripts.

Your mission is to improve the parts that matter most for live trading: model validity, no-leakage evaluation, feature quality, uncertainty estimation, auction ranking, portfolio selection, and risk-adjusted decision quality.

Key responsibilities

•

Take technical ownership of the ML, forecasting, ranking, and decision-system layer of Elcast.ai.

•

Challenge current models for leakage risk, weak assumptions, unstable features, overfitting, and backtest artifacts.

•

Improve validation and reproducibility standards so research accurately represents live-trading conditions under strict point-in-time discipline.

•

Develop, validate, and deploy forecasting, ranking, uncertainty, and portfolio-selection methods to optimize risk-adjusted trading returns.

•

Improve modularity, interfaces, tests, and guardrails where needed without unnecessary reinvention or behavioral drift.

•

Work daily with the founder, current data scientist, and ETL pipeline engineer to move the platform toward trading readiness on an aggressive timeline.

•

Help define practical standards for experimentation, model promotion, code review, monitoring, documentation, and AI-assisted development.

Required qualifications

The following are critical. Candidates who are strong only in generic modeling, notebooks, dashboards, or ETL-only work will not be a fit.

Domain and problem understanding

•

Strong ability to understand economically real prediction problems where modeling quality directly affects business outcomes.

•

Ability to quickly learn specialized market mechanics and convert them into robust forecasting, ranking, and decision logic, including hierarchical entities, timing constraints, capacity/supply-demand drivers, disruptions, and item-level ROI or risk-adjusted return. Prior energy-market, commodities, auction, or trading-domain experience is a plus, not a requirement.

•

Ability to work with hierarchical market entities such as countries, bidding zones, corridors, child corridors, auction products, and auction results without losing semantic precision.

ML and statistical requirements

•

Deep hands-on experience with modern machine learning for structured, time-dependent data, especially gradient boosting and related methods.

•

Strong command of time-aware backtesting, strict no-leakage evaluation, walk-forward validation, and model-quality metrics beyond simple point accuracy.

•

Ability to build and evaluate forecasting, ranking, and decision-support models, including learning-to-rank, top-K selection, quantile regression, uncertainty estimation, and abstention logic.

•

Understanding of financial modeling, portfolio selection, capital allocation, drawdown, false-positive cost, and risk-adjusted return.

•

Ability to design and review feature engineering at scale, including lagged, rolling, aggregated, hierarchical, and cross-source features.

Engineering requirements

•

Senior-level Python engineering ability, not just notebook-based modeling.

•

Experience with MLflow or comparable experiment-tracking/model-lineage tooling.

•

Familiarity with Pydantic-driven configuration and type-driven architecture, including validators, immutable configuration objects, and explicit interfaces.

•

Comfort with modular pipelines, strategy patterns, registries, tests, and refactoring existing code without behavioral drift.

•

Excellent SQL fluency, including schema-qualified queries, function-based pipelines, bulk operations, idempotent writes, and performance-aware data access.

•

Ability to review and improve code for correctness, maintainability, reproducibility, security, and collaboration safety.

Collaboration and execution requirements

•

Strong business orientation and willingness to work within a founder-led, high-accountability environment.

•

Low-ego, facts-first, productivity-first working style; able to improve existing work without politics or unnecessary reinvention.

•

Ability to inherit substantial prior work, collaborate respectfully with the current team, and move the system toward trading in approximately three months.

•

Ability to work productively with AI-assisted development while independently reviewing and challenging generated code.

Preferred qualifications

•

Experience in energy, commodities, quant systems, auctions, optimization, or markets where prediction quality has immediate economic consequences.

•

Experience with portfolio selection, capital allocation, optimization, or risk layers sitting on top of model outputs.

•

Experience with unit testing, integration testing, model monitoring, model registries, and strong repository hygiene.

•

Comfort with modern packaging, command-line tooling, and production-ready project structure.

•

Ability to write clear architecture notes that explain not only what the system does, but why important design decisions were made.

What the right candidate gets

•

A genuinely hard and economically meaningful ML problem.

•

A serious existing platform with MLflow lineage, web-based model review, data provenance, and substantial leverage already developed.

•

Direct daily collaboration with a founder who understands both the technical and commercial system in depth.

•

Meaningful ownership and the chance to shape the technical DNA of the company.

•

The opportunity to convert years of domain work and recent modeling breakthroughs into a durable trading capability.

Compensation, Equity and Co-Investment Opportunity

We view this as a high-ownership, company-shaping technical role and intend to compensate it accordingly.

Compensation

•

Base salary: €110,000-€135,000, depending on experience, technical depth, and expected scope of ownership.

•

Annual bonus: up to 15%, tied to defined technical and delivery milestones.

Equity

•

Initial equity grant: approximately 1.0%-1.25%.

•

Additional milestone-based equity designed to bring total potential ownership to roughly 1.5%-2.0% for an exceptional candidate who delivers at the expected level.

•

Equity structure, vesting, and milestone design will be discussed in detail during the process.

Co-Investment Opportunity

•

Subject to legal structure, internal policy, and individual eligibility, senior team members may have the opportunity to invest up to €100,000 of personal capital alongside the company’s JAO trading strategy, with no company commission.

•

Any such arrangement would be subject to formal documentation, compliance requirements, conflict-management rules, and jurisdiction-specific legal and tax treatment.

Apply

If you want to inherit serious technical work, improve it intelligently, and help build a system that can win in the market, we would like to hear from you. If your profile is strong on both code and model judgment, but you do not match every single line item exactly, you are still encouraged to apply.