Thomas K. Baldwin Curriculum Vitae

Joined Valtech via its acquisition of Kin+Carta in April 2024. Promoted to Senior Principal in September 2024.

• Served as deputy lead for a major data & analytics account comprising 50 consultants and $15M in annual client spend, contributing both strategic and tactical support.
  • Helped the business navigate client-side leadership changes by preparing technical summaries for new leaders, preserving our working relationship across two changes in CDO/CEO.
  • Supported casting scientists, engineers, and other data analytics professionals into compatible client-side roles. Maintained and expanded interview processes to evaluate candidates on highly technical criteria.
  • Helped tune communications to leaders, respond to delivery problems, and recognize outstanding work within the team.
• Rebuilt a client's data-driven store operations planning tool using linear optimization methods, as individual contributor and technical lead of the consulting team.
  • Owned design and implementation of a new time series forecasting component to predict demand at 15-minute intervals across 11,000 stores. Developed a framework to create accountability by tracking accuracy metrics (MAE, sMAE, MAPE).
  • Orchestrated forecasting and optimization jobs using Databricks Lakeflow pipelines, bridging the client's established development patterns with vendor-recommended best practices.
  • Coached a junior data scientist through development of the linear optimization component using Google OR-Tools.
• Provided internal support for the Valtech data practice.
  • Coordinated communities of practice among data science practitioners, establishing biweekly sessions on topics including segmentation, time series forecasting, and generative AI.
  • Served as a people leader for 5 data scientists and engineers.

Joined Kin+Carta via acquisition of Cascade Data Labs in 2021, following promotion to Principal in December 2020. CDL operated as a subidiary for two years and eventually became the core of K+C's Data Practice.

• Built next-generation personalization (NGP), a marketing optimization system, for a major quick service restaurant (QSR) client. NGP enables marketers to tailor offers to specific priorities: new game constructs, hybrid offer types, and custom optimization objectives.
  • Architected model training and champion/challenger evaluation using Databricks and MLflow, scheduled scoring and deployment using Prefect. Established a cadence for a model lifecycle that led to continuous improvement.
  • Worked with client-side data scientists to move Python code out of notebooks and into importable modules with test suites. Used GitHub Actions and Jenkins to build packages and deploy them to environment-separated JFrog Artifactories. Introduced a versioning scheme based on setuptools-scm to improve traceability.
  • Set standards for code quality on a 6-person team of consultants. Established patterns for code review and implemented pre-commit checks for repos on developer machines and in CI.
  • Provided individual contributor support for a subsequent initiative to automatically optimize marketing send efficacy using integer programming methods (Google OR-tools, CP-SAT).
• Managed teams of data scientists and engineers as a technical project lead.
  • Supervised a team of data scientists working on improving accuracy in demand forecasting products for a network of 10,000 QSR retail locations.
  • Led an internationally-distributed team of data engineers in modernizing store performance reporting. Increased reliability of BI pipelines by migrating them out of low-code tools like Alteryx and into Databricks, GitHub, and Azure Data Factory (ADF).
  • Managed another team of data engineers working on a green-field data product for environmental, social and governance (ESG) reporting, helping to navigate process-encumbered environments and respond to casting changes.
  • Contributed to client-facing quarterly business reviews.
  • Supported technical pre-sales scoping for a client seeking to develop a personalized recommendation system for PC game settings.
• Provided internal support for the Kin+Carta data practice during a period of rapid expansion of domestic and nearshore consulting staff.
  • Technical owner of Python data science interview and PySpark data engineering interview.
  • Co-designed skill benchmarks for promoting data scientists. Established performance review processes and served as data science skill council lead.
  • Scrappily handled pain points of an international expansion: shipping client laptops through customs and configuring Twilio bridges for 2FA texts.
• Fostered collaboration within the expanded data practice.
  • Led ceremonies to improve cross-pollination among data science projects.
  • Mentored 4 co-workers in quarterly career development sessions. Served as people leader for 3 direct reports.
  • Produced a 3-part training/workshop series orienting data practitioners to git, GitHub, and effective habits for collaboration on source code.

• Developed and deployed models for a major personalized marketing program, including: Uplift optimization trees (causal forests), Bayesian transaction count models, anomaly detection, Prophet time series forecasting.
• Contributed major development of a client's custom uplift modeling library: API design, code packaging/distribution, PySpark integration, and runtime performance optimizations to reduce cloud resource consumption by 10-100x.
• Led development of MLflow-based toolchain for automatic model training and scoring. Improved visibility into model performance and unlocked 4x faster iteration by reducing operator effort.
• Implemented simulation framework for offline testing of reinforcement learning models using Ray-based parallelism.
• Designed and conducted real-world tests to measure campaign performance and generate training data for optimization models.
• Built and maintained automatic weekly campaign reporting solution for a large customer loyalty program. Distilled terabyte-scale raw data inputs to rich interactive Tableau dashboards using Spark SQL and Airflow.
• Conducted ad hoc analyses for business stakeholders. Including dataset creation, formulating new KPIs, visualizing and presenting findings.

• Taught two sections of calculus-based physics:
  • PH 211 (F16+W17, 10 credit hours): Nature of science, mathematical representation, classical Newtonian mechanics, energy and momentum. With lab.
• Taught two courses in conceptual physics for non-majors:
  • PH 101 (F16, 4 credit hours): Data analysis, laws of motion, momentum, energy, gravity. With lab.
  • PH 103 (S17, 4 credit hours): Atomic physics, quantum mechanics, nuclear physics, special and general relativity, astrophysics. With lab.
• Participated in faculty seminar on the Scholarship of Teaching and Learning (S17). With Dennis Gilbert.

• Updated a LIDAR model for a local optics company.
• Researched and applied methods to account for pulse stretching induced by forward light scattering in seawater.
• Translated model code from MATLAB to Python.

• Created Snotelier, a site implementing automated avalanche danger summaries for climbers and backcountry skiers.
• Built dataset in PostgreSQL by scraping data from Northwest Avalanche Center website and combining with USDA snow telemetry (SNOTEL) measurements available via public API.
• Engineered features and fit a logistic regression model to predict avalanche danger using pandas and scikit-learn.
• Implemented web app based on Flask, Bootstrap, and the Google Maps API.
• Project completed in less than three weeks.

Adviser: Hailin Wang.

• Primary project: investigating interlayer exciton-correlated hole tunneling in Gallium Arsenide (GaAs) mixed-type quantum well (MTQW) structures at high magnetic fields.
• Side project: Diamond nitrogen-vacancy (NV) centers as a quantum bit candidate. Ion implantation and annealing, imaging of single emitters using homebuilt confocal microscope, manipulation of ground state spin using microwaves, excited-state spectroscopy at cryogenic temperatures, photon antibunching.
• Pump-probe spectroscopy of excitons, trions, and biexcitons in semiconductor quantum wells (GaAs, CdTe) and Bragg microcavities.
• Developed pulse shaper based on a phase-only spatial light modulator (SLM), with protection system based on BBO crystal frequency doubling.
• Two-time high-field user at National High Magnetic Field Laboratory, Tallahassee, FL. Operated 25T Split-Florida Helix at cryogenic temperatures, in both Voigt and Faraday geometries.
• Numerical modeling of NV centers (optical Bloch equation), microcavities (transfer matrix method), and valence subbands in GaAs quantum wells (Luttinger-Kohn Hamiltonian) using Python and MATLAB.
• Developed Python-based data acquisition utilities: Instrument communication library, confocal microscope graphical user interface.

• Co-instructed PHYS 155, “Physics Behind the Internet”, a course for non-physics majors, in Fall 2015. With Eric Corwin.

• Piloted FOSS inquiry-based science curriculum in rural school districts (Boardman, Pendleton), grades 4-6. Daily team teaching during four 2-week stays as scientist in residence.
• Guest scientist, Crow Middle/High School (2011). Supervised students preparing for EWEB solar car challenge. Developed curriculum on solar spectrum and filter functions.
• Developed bicycle smoothie machine for Charlemagne Fox Hollow Elementary School, Eugene, OR. With Carrie Grabowski (2011, 2012, 2013, 2015).

• Instructed three sections of tutorial recitation for PHYS 201-203, a three-term course in algebra-based general physics (2009-2010).
• Grader for astronomy courses (2008-2009).
• Tutor in the physics drop-in help center.

Adviser: Kevin Silverman.

• Photoluminescence and annealing of self-assembled quantum dots.
• Translated data-acquisition programs from MATLAB to Python.

Adviser: Pavel Kabos.

• Finite-element modeling of a near-field microwave microscope using HFSS software.
• Characterization of carbon nanotube samples using this microscope.

• Helped students with calculus and linear algebra in the college math lab.

• Led weeklong first-year orientation (2005, 2007), weekend trips, and indoor rock climbing sessions.
• Activities Manager, Fall 2006.
• Climbing Manager, Fall 2007.

Adviser: Carlos Ruiz.

• Surveyed activation products using gamma spectroscopy.
• Fitting of neutron-yield data from nuclear fusion experiments.

Helped coordinate a monthly NumFOCUS-sponsored meetup series.

Support-level member.

Support-level member.

• Helped organize practice sessions in backcountry navigation, snow travel, and low-angle rope systems.
• Served on board of directors as training coordinator in 2017.