09.10.2026 aktualisiert
Premiumkunde
100 % verfügbarApplied Scientist & Machine Learning Engineer
Augsburg, Deutschland
Nur Remote
BA Honours Mathematics, McGill University (2014-2019)Über mich
My background is in Mathematics. Most recently I worked at One Auction View as the primary data scientist owning the whole pipeline from ingesting blobs into delta storage lake, all the way to running automated pipelines through Terraform. I wrote a generative pipeline, based on hazard kernels. Etc.
Skills
Azure DatabricksAzure DevOpsHidden Markov ModelPythonEreigniszeitanalyseLineare AlgebraMachine LearningWahrscheinlichkeitstheorieNumPyAzure Machine LearningAzure Data LakeSQLWahrscheinlichkeit und StatistikStatistikenBestärkendes Lernen
Machine Learning & AI Engineering
Expertise in JAX (GPU/TPU), transformers implemented from the paper, LLM agents via LangChain, RAG, and computer vision. End-to-end ownership of ML pipelines from problem definition through modelling, validation, and production deployment.
Statistical Modelling & Analysis
Deep knowledge of survival and time-to-event analysis, Bayesian inference, hierarchical and empirical-Bayes models, variance decomposition (ANOVA), causal identification, and experiment design.
Data Engineering & Infrastructure
Proficiency with Databricks, Spark, Terraform, Postgres, FastAPI, SQL warehousing, geospatial tooling (H3), and Git for building and maintaining scalable data infrastructure.
Programming Languages
Proficiency in Python, SQL, Haskell, Julia, and JavaScript, with experience building production systems and research prototypes across multiple paradigms.
Forecasting Systems
Design and deployment of large-scale forecasting systems, including a vehicle-listing scoring pipeline processing tens of millions of US listings daily on Databricks.
Computer Vision
Application of semantic segmentation and photogrammetric methods to real-world problems, including geocoding structures from address data and deriving structural measurements from imagery.
Active Inference & Multi-Agent Systems
Research and implementation of networked active inference using variational free energy in multi-agent POMDPs, including formal proofs and simulation verification.
Technical Writing & Teaching
Authoring of pilot articles on first-order optimisation methods for e-learning platforms, and tutoring in first-order logic, natural deduction, and denotational semantics in Haskell.
Expertise in JAX (GPU/TPU), transformers implemented from the paper, LLM agents via LangChain, RAG, and computer vision. End-to-end ownership of ML pipelines from problem definition through modelling, validation, and production deployment.
Statistical Modelling & Analysis
Deep knowledge of survival and time-to-event analysis, Bayesian inference, hierarchical and empirical-Bayes models, variance decomposition (ANOVA), causal identification, and experiment design.
Data Engineering & Infrastructure
Proficiency with Databricks, Spark, Terraform, Postgres, FastAPI, SQL warehousing, geospatial tooling (H3), and Git for building and maintaining scalable data infrastructure.
Programming Languages
Proficiency in Python, SQL, Haskell, Julia, and JavaScript, with experience building production systems and research prototypes across multiple paradigms.
Forecasting Systems
Design and deployment of large-scale forecasting systems, including a vehicle-listing scoring pipeline processing tens of millions of US listings daily on Databricks.
Computer Vision
Application of semantic segmentation and photogrammetric methods to real-world problems, including geocoding structures from address data and deriving structural measurements from imagery.
Active Inference & Multi-Agent Systems
Research and implementation of networked active inference using variational free energy in multi-agent POMDPs, including formal proofs and simulation verification.
Technical Writing & Teaching
Authoring of pilot articles on first-order optimisation methods for e-learning platforms, and tutoring in first-order logic, natural deduction, and denotational semantics in Haskell.
Erfahrung
Berufserfahrung5 - 10 Jahre
Freelancer seit2023
Sprachen
DeutschGrundkenntnisseEnglischMutterspracheFranzösischgutRumänischverhandlungssicher
Projekthistorie
Sole data scientist for RetailVision, OAV's used-car analytics product. Owned the full pipeline from problem definition to production. Identified and corrected systematic bias in the incumbent sales estimator using survival analysis, measured model-free predictability ceilings, validated inferred labels against external ground truth, prototyped and rejected a hierarchical Bayesian hazard model, and shipped a simple per-segment exponential-hazard MLE running autonomously on Databricks over tens of millions of US listings.
Applied engineering contracts including computer vision for disaster mitigation (geocoding structures, semantic segmentation, photogrammetric door height derivation) and technical writing on first-order optimisation methods for an e-learning platform, plus tutoring in logic and Haskell.
Research on networked active inference using variational free energy in multi-agent POMDPs. Formalised agent hidden states, proved free energy decomposes additively over neighbours under channel independence, and implemented and verified the framework in Python via simulation over chains, stars, and grids.