Machine Learning Engineer (Time-Series & Decision Systems)
Location:Porto (On-site)
Company:Sybilion
AtSybilion, we are building thedecision layer for industrial markets.
Industrial companies operate in environments shaped by thousands of external signals: commodity prices, trade flows, energy markets, weather, geopolitics, logistics constraints, and shifting demand.
Most decisions today are still made using spreadsheets and intuition.
Sybilion changes that.
Our platform discovers the signals that truly matter acrossbillions of time-series data pointsand turns them intodecision-ready intelligence.
Procurement teams decide when to buy.
Commercial teams decide how to price.
Supply chains decide how to position inventory.
We are not building dashboards.
We are building theworld model industrial companies use to make decisions under uncertainty.
To do this, we are looking foroutlier ML engineerswho want to work on one of the hardest problems in applied machine learning:understanding complex real-world systems through time-series modelling.
The Role
You will work directly on thecore modelling systems behind Sybilion's decision layer.
This means building models that discover relationships across massive time-series datasets and turn them intoreliable forecasts, causal insights, and decision signals.
You will work closely with the founders and product team in ahigh-ownership, high-impact environment.
What You'll Work On
Time-Series Modelling at ScaleBuild models that understand and forecast complex market systems.
Examples include:
• Commodity and chemical price dynamics
• Supply-chain demand signals
• Trade flow shifts and macro indicators
• Energy and logistics cost dynamics
• Production and consumption cycles
You will design and evaluate models such as:
• classical forecasting models (ARIMA, ETS, state-space)
• modern ML models for time-series prediction
• multivariate signal discovery models
• regime detection and structural break modelling
• probabilistic forecasting and uncertainty estimation
Signal DiscoveryOne of Sybilion's core problems is discoveringwhich signals actually matter.
You will work on systems that:
• search large time-series spaces for predictive relationships
• identify causal drivers
• detect regime changes and structural breaks
• filter noise from true market signals
Decision-Grade ForecastingOur models do not exist for research papers — they exist to drivereal decisions.
You will help translate models into:
• decision thresholds
• uncertainty ranges
• scenario simulations
• early warning signals
Your work will influencemillions of euros in procurement and pricing decisions.
Production SystemsYou will help turn modelling approaches intorobust production pipelines, including:
• feature pipelines for large time-series datasets
• scalable modelling workflows
• evaluation and backtesting frameworks
• automated model monitoring
Who We're Looking ForWe are looking foroutliers.
People who move faster, think deeper, and build better systems than the average engineer.
You may recognise yourself if you:
• obsess over understanding real systems
• enjoy modelling messy real-world data
• care about correctness and robustness
• like solving problems that are not well defined
Must Have• Strong Python (NumPy, pandas, PyTorch/JAX/Scikit-learn or similar)
• Experience buildingtime-series models
• Deep curiosity about real-world systems
• Ability to design rigorous evaluation and backtesting frameworks
• Comfort working with large datasets and imperfect data
• Strong problem-solving ability
Strongly Valued• Experience with forecasting methods (ARIMA, ETS, Prophet, state-space models)
• Multivariate or causal modelling
• Probabilistic forecasting
• Experience with large-scale time-series datasets
• SQL / data pipelines / cloud environments
• Experience in commodities, macroeconomics, or supply chains
What Success Looks Like (First 6 Months)• You build models that materially improve forecasting accuracy in at least one market domain
• You understand and advise customers
• You contribute to Sybilion'ssignal discovery engine
• You design robust evaluation and backtesting systems
• Your models are used by customers in real decision workflows
Why Join SybilionWe are building a company at the intersection of:
• machine learning
• economics
• complex systems
• industrial decision-making
This is not a typical SaaS product.
It is an attempt to build thedecision infrastructure for real-world markets.
If that excites you, we would love to talk.