Yaolin Ge · Trondheim, Norway
I build statistical machine learning that has to work outside the notebook.
PhD statistician turned production engineer. At Sandvik Coromant I take anomaly detection on multi-sensor machining data from a Python study to shipped C#, and build the GPU visualisation engineers use to trust it. Before that: robots that decided where in a fjord to measure next.
- 11/11machining faults caught by a shipped detector, with zero false alarms
- 3 yrstaking ML into a production .NET and TypeScript product
- 5peer-reviewed papers on adaptive sampling and planning
- 2.5 hautonomous underwater mission run by my planner
What I work on
All skills →- Deep · daily work since 2023 Time-series ML & anomaly detection Finding the moment something goes wrong in multi-sensor streams, and proving the detector is right before anyone relies on it.
- Applied · notebook to product ML in production & MLOps Getting a model out of the notebook and keeping it honest afterwards: export, inference inside a .NET product, pipelines, tracking and monitoring.
- PhD, Applied Statistics · NTNU Bayesian & spatial statistics Models that know how uncertain they are, and decisions that use that uncertainty: where to sample next, when to replan, what to believe.
- Production · 2.5 years GPU visualisation & interactive UIs Making very large, very fast sensor data explorable: WebGL2 rendering, real-time views and comparison tools that engineers use at the machine.
- Production · identity first Azure cloud & DevOps Cloud storage and services designed around least privilege, provisioned by script and delivered by pipeline.
- ~560 commits · 30 merged PRs Software engineering Two and a half years on a production .NET and TypeScript product: code review, tests, CI/CD and refactoring that keeps a large codebase moving.
Selected projects
All projects →- 2026 A fault detector the shop floor can trust A causal streaming detector for cutting-insert faults, taken from a Python study to shipped C#. It caught every labelled fault without a single false alarm.
- 2026 From labelled cuts to C# constants An Azure ML pipeline that trains candidate detectors in parallel, picks a champion against the challengers, and exports the winner straight into production code.
- 2025 Thousands of sensor traces, drawn on the GPU A legacy SVG chart re-architected into a WebGL2 and Canvas engine that overlays 1000+ series per plot and opens 14-hour recordings, with comparison tools built on top.
- 2025–26 Comparing a cut with a good one, while it happens A live view that streams machining data against a known-good reference curve, so an operator sees a deviation during the cut instead of in the report afterwards.
- 2020–23 Teaching an underwater robot where to look Bayesian models and a planner that ran onboard an autonomous underwater vehicle in Trondheimsfjord, steering it along a river-plume front instead of surveying evenly.
- 2022–25 Planning a long way ahead, around obstacles An RRT*-based planner that looks beyond the next waypoint: it trades information against distance over a cost valley while respecting obstacles and a time budget.
The path so far
- 2018 Dual MSc, marine technology KTH and NTNU. Vehicle dynamics, control, and estimating where a marine robot is.
- 2019 Research intern, Peking University Camera to 3-D pose to a humanoid robot, live on stage.
- 2020 PhD, applied statistics, NTNU Adaptive ocean sampling with SINTEF Ocean and LSTS, run onboard in the fjord.
- 2023 ML & software engineer, Sandvik Coromant Sensorized cutting tools: detection, visual analytics, Azure.
- Now Production fault detection A detector the shop floor can trust, and what I'm thinking about.
Recent writing
- MASCOT — teaching an underwater robot where to look Adaptive sampling for AUVs, from the Gaussian random field to a planner running onboard in Trondheim fjord
- Regex expression reference Cheatsheet for regex expression
- Amdahl's Law Math derivation of Amdahl's Law
- Travelling in Shanghai Holidays in Shanghai
Publications
All five →- RRT*-enhanced long-horizon path planning for AUV adaptive sampling using a cost valley Knowledge-Based Systems, 2025
- 3-D Adaptive AUV Sampling for Classification of Water Masses IEEE Journal of Oceanic Engineering, 2023 · doi · pdf
- Long-Horizon Informative Path Planning with Obstacles and Time Constraints IFAC-PapersOnLine, 2022 · doi · pdf