Skip to main content

Resume (download as 1-page PDF file)

 

Andrew Ganse, PhD
andrew@ganse.org   |   github.com/aganse   |   research.ganse.org   |   www.linkedin.com/in/andrew-ganse

 

SUMMARY

Hands-on senior data scientist/ML engineer and tech lead: 8+ years shipping production ML and evaluation systems and 12+ years of statistical modeling, uncertainty quantification, and model evaluation on large, noisy, real-world data. Skilled at turning ambiguous problems into measurable metrics and production models. I write the code and run the analyses end to end, from data discovery and pipelines through validation, training, deployment, and monitoring. Startup veteran who delivers customer pilots and works closely with engineering and product.

 

PROFESSIONAL EXPERIENCE

Senior Data Scientist — ThruWave Inc., Seattle WA · Jul 2020 – Mar 2026 (company closed)
Data science + machine learning for automated final QC of packaged products in warehouses.

  • Designed and built the company’s ML and evaluation platforms - experiment tracking, model training, evaluation, deployment, and monitoring - for computer-vision anomaly and object detection in mmWave, depth, and RGB images in production.
  • Built production return-fraud detection: per-product models flagging returned boxes whose contents deviated from the expected population, catching adversarial substitutions (e.g. a brick for a mini-PC matched to the original on size and weight), under severe class imbalance and false-positive cost constraints.
  • Delivered 10+ production pilots at major corporations and government agencies, shipping 16+ models, converting multiple deployments into paying customers.
  • Owned data engineering end to end: built and maintained a 170TB image corpus backing training-set curation; reprocessed and recalibrated the full corpus on a temporary parallel EC2 fleet.
  • Led and mentored a 4-member data science team and set roadmaps with product/engineering while remaining an individual contributor; improved delivery velocity ~50% via AI coding agents, CI/CD, and standardized ML workflows.
  • Applied transformer-based vision-language models (OpenCLIP) for zero-shot and centroid-based classification and evaluation, building the embedding and similarity-search tooling behind them.
  • Lead inventor on granted patent for high-throughput anomaly detection in 3D millimeter-wave imaging systems.
  • Tech: Python, SQL/PostgreSQL, OpenCLIP, Scikit-learn/Pandas, PyTorch/TensorFlow, OpenCV, MLflow, AWS, Docker, dbt, Git.

Data Scientist — Echodyne Corporation, Kirkland WA · Dec 2017 – Jul 2020
Data science + machine learning on meta-materials tracking radar.

  • Built and scaled the company’s field-test data, labeling, and model performance evaluation platform, automating time-series processing, dashboards, and reporting across 100+ live tests, raising evaluation cadence from monthly to weekly.
  • Built the company’s first ML-based object classification capability for a production system, driving increased customer adoption, expanded implementation in follow-on versions, and influencing 10+ feature releases.
  • Designed matched-condition field trials to measure model performance changes from release to release.
  • Developed ML for autonomous automotive perception and drone classification using radar, GPS, RGB camera.
  • Tech: Python, SQL/PostgreSQL, Scikit-learn/Pandas, TensorFlow/Keras, Airflow, MLflow, Docker, Matlab.

Principal Scientist — Anseres Research & Technology LLC · Sep 2016 – Dec 2017
Federal scientific consulting in space/defense.

  • Technical evaluation and advisory work for defense and space R&D programs, supporting federal contracts across sensing, signal processing, and estimation systems.
  • Published radio-science gravimetry research at conference with NASA collaborators.
  • Led development of SBIR proposal on deep learning approaches for clutter reduction in sonar systems with university collaborators.

Data Scientist — Spare5 · Jan 2016 – June 2016.
Crowdsourced data labeling with machine-learning-driven quality control.

  • Developed machine learning models for data quality and user reputation evaluation on Spare5’s intelligent-crowdsourcing platform for data labeling. Word2vec, SQL/PostgreSQL, R/Rserve, Python/Pandas.
  • Produced model that resulted in 8% lift in search-traffic to a customer’s data in first few months of job.

Senior Research Physicist, Applied Physics Lab, Univ. of WA · Apr 1999 – Nov 2015
Applied research and statistical modeling in ocean-acoustic, electromagnetic, and gravimetric sensing domains.

  • Researched wave propagation in random media, inverse problems, and sensors: acoustic oceanography, seafloor seismo-acoustic propagation, controlled-source electromagnetic sensing, radar scattering, spacecraft radio-science gravimetry.
  • Solved nonlinear regression, inversion, optimization, tracking, and signal processing problems using Python, Matlab, Octave, C, Java, Fortran, Linux shell scripting; administered Linux clusters.
  • Built Markov Chain Monte Carlo based Bayesian inversion of acoustic data on a Linux cluster.
  • Developed Kalman nonlinear smoothing/tracking and parameter estimation algorithms to acoustically track a 5km(!) long vertical hydrophone array for our ocean acoustic experiment.
  • Designed research experiments, presented results at conferences, wrote proposals and reports to sponsors, interfaced with sponsors, wrote and reviewed research papers.
  • Led system engineering and field testing for an experimental “towed CTD chain” cabled instrument containing 90 sensors, managed teams of 2-6 others at a time in testing and operation of the system. Managed two students in statistical programming projects.

 

EDUCATION

Ph.D. Geophysics, University of Washington: statistical inference, parameter estimation, inverse problems.
B.S. Electrical Engineering, University of Washington: nonlinear optimization and signal processing.

 

SELECTED SIDE PROJECTS

(click icons to see GitHub repo and article in this website)

RAG_MLflow_eval : Langchain-based RAG evaluation with LLM-as-judge, logging datasets and traces to MLflow.
Semantic_SQL_examples : Semantic SQL using pgvector, ordering review comments along semantic axes.
Flow_models : Flow-based invertible neural networks implemented in TF Probability for synthetic image generation.