Primarily focused on internal customer-driven development of data systems for malware labelling, and ML workflows. Extensive user story development, collaboration with platform teams, and alignment with corporate priorities. Core tools include AWS (S3, EMR Spark + Trino, Athena / Glue), Hive + Iceberg tables, Apache Superset, Grafana, PostgreSQL, Kafka, HTTP microservices, GNU Make, Docker (with a sprinkling of k8s and terraform), documents and diagrams of all types, and - sadly - Jira.

Strategic planning across engineering and product teams on a complex distributed data science product. Improved scalability and enabled headless compute with data model and APIs for managing Jupyter compute kernels. Developed observable, testable, async code using FastAPI and Pydantic to manage and publish compute state. Assisted backend and frontend engineers with context, extensive pair programming, and site reliability.

Supported a startup during a job search, leading data science efforts with automotive machine vision and telematics. Developed reusable interactive dashboards, resulting in timely customer delight in a sales cycle for a major client. Provided mentorship to the junior data science team, focusing on improved reusability, testability, and interactivity.

Implemented proofs-of-concept in Rasa and Azure Bot Framework for a stress reduction intervention. Implemented data collection practices, machine vision pipelines, and interactive web-based visualizations for a musculoskeletal (MSK) health mobile app. Created a structured knowledge framework for an MSK domain expert. Cut lead time to implement machine vision exercise prescriptions by three months from 6 months. Bridged collaboration between differently skilled teams and contributed core algorithms and data for a mobile product. Evaluated, hired, and managed multiple ML and MLops teams. Managed work across Jira boards, Clubhouse, and Google Docs to keep individual contributors focused and productive. Set up HR, taxes, GitHub, Azure, and Amazon Web Services (AWS).

Enabling transparent, reproducible, collaborative, and open science by automating Docker, Git, and related approaches to portability. Coordinating research to improve education. Creating training materials and presenting internationally and at the World Bank.

倉楽 (Kuraraku) is a Maryland LLC that literally translates as "warehouse of ease." I provided research, design, and educational services with a focus on youth, learning, and human development. The primary focus was a contract with a Japanese educational company to change the way the world learns via cutting edge approaches to measurement and analysis. The initial focus employed survey and cognitive testing methods. This work culminated in a classroom machine vision system that is still actively developed and used.

Develop and assess interventions for childhood developmental challenges like ADHD and Autism. Focus on "mindful movement" interventions. Working in the Center for Neurocognitive and Imaging Research (CNIR).

Collaborating on the Moore-Sloan data science initiative, in particular working on tools to enable research using mobile devices.

Bridge the gulf between cutting edge scientific computing and graduate education in the social sciences. Led multiple internal and external training initiatives. Mentored grad student workers.

Working on systems to measure social and environmental data. Collaborating on projects in the context of my Hacking Measurement class at UC Berkeley.

Teaching the practical Machine Learning course.

Making sure we actually know how to change our own carbon footprint. Applying the scientific method to the rather pressing social problem of climate change and human behavior.

We do web hosting and design using the Zope-based content management system, Silva

Grad student in Rich Ivry's CognAc lab and Michael Ranney's Reasoning group. I like using complicated statistical models to study close to real-world phenomena. As you might imagine, those phenomena are kept as simple as possible.

Primarily focused on analysis and alternative implementations of core web page classification algorithms. Worked closely with human labeling team to develop a full-stack data-to-product pipeline that worked.

Prototype database system / web interface for brain imaging research.