Sybilion builds AI-driven market forecasting for process industries (chemicals, packaging, pulp & paper, textiles, broader manufacturing). We help procurement, supply-chain, and commercial teams make confident buy/sell decisions with clear, defensible forecasting signals.
We're hiring a model-focused Consultant who combines consulting rigor with hands-on data science. You'll be deeply involved in forecast design, validation, and deployment into client workflows, while also supporting discovery, PoCs, and executive communication. You'll work closely with the founders in a high-impact role based in Porto.
The mission
Turn messy market + operational data into forecasting models clients trust - and translate model outputs into decisions that move cost, inventory, service level, and margin.
What you'll do:
Build & validate forecasting models
* Design forecasting approaches for prices, demand, lead times, consumption, inventory risk, and volatility depending on the use case.
* Run EDA, feature construction, and baseline benchmarking (e.g., naive/seasonal, ETS/ARIMA/Prophet, ML models where appropriate).
* Own model evaluation and sanity checks: backtesting, leakage checks, regime shifts, outliers, structural breaks, and "does this make business sense?"
* Define and standardise metrics and reporting (MAPE/sMAPE/WAPE, bias, coverage, confidence bands, error by horizon/segment).
Operationalise models into decision workflows
* Convert model outputs into decision-ready artefacts: recommended actions, risk flags, thresholds, what-changed narratives, and "so what" implications.
* Help shape model outputs into templates, dashboards, and playbooks used by procurement/S&OP/pricing teams.
* Improve repeatability: contribute to internal model libraries, notebooks, evaluation harnesses, and delivery templates.
Model-led PoCs and client delivery
* Partner with sales on discovery to frame forecasting hypotheses, required data, and what "success" looks like.
* Lead PoCs end-to-end: data intake → modelling → backtest → insights → exec readout.
* Run weekly client cadence: progress, risks, stakeholder alignment, and value tracking.
* Handle live Q&A with credibility: explain why the model says what it says, where it's uncertain, and what we'll do next.
Who we're looking for
You're model-strong (not just "data literate")
* You understand the forecasting problem space (time-series, seasonality, volatility, segmentation, horizons) and can choose sensible approaches.
* You can explain trade-offs clearly: accuracy vs interpretability, stability vs responsiveness, model complexity vs maintainability.
You're credible with enterprise clients
* Polished, reliable, precise, high follow-through — our clients are conservative and detail-oriented.
* You can align stakeholders around what the model will (and won't) do, and create champions.
Must-haves
* 3–7 years in management consulting (MBB/Big 4/boutique) or client-facing data/analytics consulting where modelling was central.
* Strong practical Python (pandas/NumPy; notebooks; building/adjusting modelling pipelines; comfortable with time-series basics).
* Experience working with forecasting evaluation and backtesting; can diagnose why forecasts fail.
* Business fluency in procurement, supply chain, pricing, or S&OP (enough to connect model outputs to decisions).
* Executive communication: can produce clear readouts that defend the model and drive action.
* English fluency; based in / willing to relocate to Porto (hybrid).
Nice to have
* Process industry exposure: chemicals, packaging, pulp & paper, textiles, or adjacent manufacturing.
* Forecasting toolset familiarity: ARIMA/ETS/Prophet, gradient boosting, causal/external regressors, hierarchical forecasting.
* SQL; Snowflake/BigQuery; dbt; Jupyter; basic cloud (AWS/GCP).
* Strong metrics instincts: bias, calibration/coverage, error decomposition, stability over time.
* Portuguese, German, or Spanish would be advantageous.
Compensation & benefits
* DOE + equity (relocation support available)
* Performance bonus
* Learning budget, modern hardware, conference travel
* Fast path to increased responsibility (e.g., Engagement Manager / Model Lead)
What success looks like (first 3–6 months)
* You ship at least one PoC where the client trusts the model's logic and adopts outputs in a live cadence.
* You standardise a repeatable modelling + evaluation workflow (baselines → backtest → reporting).
* You materially improve forecast quality/consistency in one segment (accuracy and trust).