Ich helfe Industrie- und Pharmaunternehmen, Prognosefehler zu reduzieren und statistische Modelle in Produktion zu bringen. Promotion in Mathematik.
Industrieunternehmen verschwenden Millionen durch falsche Sicherheitsbestände
Ich baue Prognosesysteme, die sich bei über 500.000 SKUs bei Unternehmen wie Kärcher und Festo bewährt haben — mit 25–35 % weniger Prognosefehlern und proaktiver Planung statt Feuerwehreinsätzen.
Datenqualitätsprobleme zerstören das Vertrauen in Ihre Zahlen
Ich schaffe automatisiertes Monitoring, das Anomalien erkennt, bevor sie Entscheidungsträger erreichen — damit Ihre Supply-Chain- und Finanzteams den Daten vertrauen können.
Ihr R/Python-Prototyp verlässt nie den Laptop des Data Scientists
Ich verwandle Notebook-Prototypen in produktionsreife Rust/WASM-Systeme, auf die Ihr Team sich täglich verlassen kann — schnell, getestet und mit CI/CD deployed.
Pharmaunternehmen brauchen GMP-validierte Methoden mit Audit-Trails
Ich liefere validierte statistische Methoden mit lückenloser Dokumentation, die Regulierungsbehörden bei Boehringer Ingelheim, Schott und vergleichbaren Unternehmen überzeugen.
From demand planning to financial forecasting — automated ML pipelines that learn and improve. Bayesian inference, hierarchical models, and causal methods grounded in a PhD in Mathematics. Proven across 500k+ SKUs.
I ship your statistical models as fast, reliable Rust/WASM systems — not notebook prototypes. Containerized, CI/CD-deployed, and monitored in production on AWS, Azure, or GCloud.
Consolidating fragmented data sources into a single source of truth — with automated quality monitoring, anomaly detection, and interactive dashboards that translate analytics into action.
Delivering measurable impact — reduced forecast errors, automated pipelines, and data-driven decisions across pharma, defense, energy, manufacturing, and finance.
Consulting a manufacturing company on demand forecasting in Kinaxis Maestro — selecting forecast metrics and levels, evaluating forecast quality, and maximizing the platform's forecasting capabilities.
Rigorous statistical analysis of spectral data in GMP-validated CMC environments for pharmaceutical manufacturing.
Automated ML pipeline for Order Intake, Revenue, and Cash Flow prediction, used by the controlling department.
Automated data quality pipeline validating input data before it feeds into the forecasting engine, with reporting on AWS S3.
Production ML pipeline on AWS SageMaker for SKU-level demand forecasting with a clear data pipeline enabling the local data scientist to run reproducible experiments.
Scheduled monthly demand planning pipeline running twice per cycle — one run for APO data and one for Kinaxis data — embedded in a strict demand planning workflow.
Specialised forecasting engine for spare parts, handling intermittent demand patterns and optimising safety stock levels to minimise stockouts and improve equipment uptime.
ETL consolidation of MS Dynamics 365 data sources into a unified risk scoring engine for Purchasing and Sales exposure assessment.
Rust-to-WebAssembly compiled forecasting engine embedded in Google Sheets, used by controlling for end-of-month revenue forecasting across subsidiaries.
Statistical strategy for BioPharma CMC — rigorous statistical analysis, process validation, and equivalence testing under GMP.
Containerized predictive analytics pipeline on MS Azure forecasting next-order dates via feature-engineered customer and territory models on automated weekly/monthly schedules.
Statistical time-series models in R predicting customer order windows, with dockerized Quarto reporting pipelines deployed on Azure Cloud.
Hybrid predictive maintenance combining photovoltaic generation models with statistical time-series forecasting and R Shiny monitoring dashboards.
Established Data Science practice — Credit Risk scoring models and Insurance Pricing engines on MS Azure with team mentoring and agile integration.
Distributed forecasting platform on Azure Databricks processing smart meter readings with Spark MLlib model training.
High-performance R/C++ package implementing Mahalanobis distance and Isolation Forest methods for multivariate outlier detection in global contract data.
A Rust-native DuckDB extension providing a complete time-series forecasting toolkit via SQL. Integrates 32 models including AutoARIMA, AutoETS, TBATS, MSTL, and intermittent demand methods (Croston, ADIDA, IMAPA). Supports hierarchical time series, expanding/sliding window cross-validation, conformal prediction intervals, changepoint detection, and 76+ tsfresh-compatible feature extraction functions with native DuckDB parallelization.
-- Forecast 10,000 products in one querySELECT * FROM ts_forecast_by( 'sales', item_id, date, quantity, 'AutoARIMA', 12, '1M', MAP{'seasonal_period': '12'});I'm Simon Müller — a mathematician turned systems engineer with over a decade of experience helping companies make better decisions through data.
After completing my PhD in Mathematics, I spent years working at the intersection of rigorous statistics and production software — first in academia, then in consulting for some of Europe's largest manufacturers, pharma companies, and financial institutions.
What sets me apart: I don't just build models — I ship them. My clients get production systems they can depend on, not prototypes that need another team to productionize. Whether that's a Bayesian forecasting engine running on AWS SageMaker or a Rust library compiled to WebAssembly running in the browser.
Based in Germany. Available for remote and on-site engagements across Europe.
Herausforderung bei Prognosen, Datenqualität oder statistischen Modellen? Lassen Sie uns sprechen — von kurzen Beratungseinsätzen bis zur vollständigen Systemimplementierung.
Wir verwenden Cookies für Analytics (Google Analytics), um diese Website zu verbessern. Ohne Ihre Zustimmung werden keine Daten erhoben.