Skip to content

Bradley Pfeil

Adelaide, SA · · GitHub · LinkedIn

Experience

Streamline (Lunio) — Data Scientist
2025 – Current
  • Worked with a team to build a dynamic pricing platform for enterprise clients that surfaces revenue opportunities through demand modelling and optimisation.
  • Modelled transition probabilities with a two-stage approach (hurdle classifier + gradient-boosted regressor with uncertainty estimates) and jointly developed a Markov Decision Process solver that recommends optimal prices and forecasts revenue.
  • Estimated price response with causal inference and added a counterfactual data pipeline so models generalise beyond observed price points.
  • Currently developing a deep Q-learning agent (PyTorch) that treats commodity-storage pricing as a sequential control problem.
  • Shipped models to production on AWS with CI/CD pipelines; refactored bottlenecks to halve inference time.
  • Cut monthly AWS spend by 20%+ via a dedicated cost-monitoring framework.
  • Worked closely with clients to explain results, shape reward functions, and keep recommendations grounded in the business; started a paper-reading group.
Aurizn — Intern / Graduate Data Scientist
2023 – 2025
  • Built LLM-based recommender systems consuming Shapley values to explain their decisions in terms non-technical stakeholders could act on.
  • Worked with the team to build client-requested features and model improvements for a dynamic pricing system, including data ingestion and automated retraining that kept accuracy steady as the data shifted and cut manual upkeep.

Education

The University of Adelaide
2022 – 2024
Bachelor of Computer Science (Advanced)

Languages & tools

Proficient
PythonSQLGBTPolarsPyTorchAWSDockerGitLinux
Experienced
PySparkRust

Additional

  • Scrum Master Certificate (2025)