Radically Different Data Science provides consultation services including: analytics and machine learning, market research, technical design, implementation, developer relations, leadership, hiring, mentorship, career coaching, and management.https://officehours.com/samantha-zeitlin

Supervised and mentored 7 engineers split into 2 areas: MLOps (endpoint malware detection and infrastructure for that), and AppliedML (Entity Analytics in the Elastic Stack)

• Oversaw productionizing the malware model and lists for faster updates and responses to customer feedback, including adding tests and CICD. • Refactored and containerized the Labelmaker system for processing 3rd-party data on malware samples, including making it scalable, adding tests, and CICD. This is now integrated into the Protections Portal, the central source of truth for the Protections team.

Built 3 end-to-end prototypes of machine learning systems for anomaly detection. Managed the data team (3 engineers). Tech Stack: GCP including GKE (kubernetes), GKB (docker builds), GCS (S3-equivalent storage), PubSub (streaming), BigQuery (data warehouse), Firestore (for model data). Python. Pachyderm (self-hosted on kubernetes with terraform). Slack webhook (for posting model results). Looker (dashboards). Weights and Biases (ML monitoring).

• Founded data science team and built data infrastructure for game data (Guns, Cars, Zombies!). • Sourced, recruited, and managed a team of four data scientist-engineers. Advised Product and Engineering on troubleshooting existing products. Supervised and built ETL pipelines for data from Apple, Google, Flurry. Identified cases of cheating using internal databases (MySQL). Stack: Python using Docker to deploy with Pachyderm in Kubernetes on AWS, set up and managed the Redshift data warehouse, and supervised building dashboards in Looker.

• Founded data science team and built data infrastructure for an AI-powered social media mobile app. • Sourced, recruited, and managed a team of four data scientist-engineers. Advised Product and Engineering on hiring, designed metrics for new data-driven features as well as troubleshooting existing products. Designed and prototyped ML features, A/B testing system.Supervised and built ETL pipelines for data from Apple, Google, Localytics, and internal databases (Postgres). >15M rows of data daily with automated pipelines. ~50 users of >24 dashboards for critical business decisions.Stack: Python and Pyspark using Docker to deploy with Pachyderm in Kubernetes on AWS, set up and managed the Redshift data warehouse, and supervised building dashboards in Looker.

Data scientist working with the Yahoo BrightRoll team (video ad exchange) (2016-2017) and the AOL ONE Video team (2017-2018).The stack: python (pandas, matplotlib, seaborn, numpy, scikit-learn, Airflow) on AWS (s3, Redshift, Spark). We also used Looker extensively for data sharing across teams. Our team had direct revenue impact. We worked directly with the Product, Engineering, and business teams to optimize supply (inventory quality) and monetization.

Data pipelines from postgres using django and pandas. File parsing, time series, pub/sub patterns, dynamic naming, data comparison and extrapolation. Increased test coverage and updated legacy code. Built a Slackbot.

• Market Intelligence related to Healthcare. • Data Science and Visualization related to Energy. • pandas• python• matplotlib, seaborn• Google maps• django

High-content imaging and statistical analysis. Assay development and optimization for target and chemistry platform validation

High-throughput screening (HTS) and statistical analysis. High-content imaging screens conducted in collaboration with Peter Walter's lab, Kevan Shokat's lab, and Genentech, studying the unfolded protein response (UPR). Addtionally, helped develop a screen for compounds that promote expansion of hematopoetic stem cells, in collaboration with Andy Leavitt's lab, and GE. Screened a panel of human cancer cell lines for sensitivity to loss of a putative oncogene, using siRNA (collaboration with Genentech).

BME50A is an introductory-level course in cell and molecular biology for engineering students. For this class, I was in charge of 150 students. I supervised two teaching assistants, who ran recitation sections and helped with grading homework, quizzes, and exams. I gave two lectures a week, 1-2 hours each, met with students during weekly office hours, posted assignments and grading rubrics to the class websites, and answered numerous student emails.

Postdoctoral FellowFull-timeDates Employed May 2002 – Feb 2010Independently defined my own project, identified collaborators, developed methods, and collaborated to develop custom software for big data analysis (CellFinder). Collaborated to use a state-of-the-art robotic laser system, which helped identify a DNA damage-dependent epigenetic mechanism for chromatin assembly at centromeres. Multiple first-author and corresponding-author publications. Research techniques: Xenopus egg extracts, protein purification, antibody generation and purification, Western blotting, siRNA, transient transfection, clonal selection, real-time quantitative PCR, recombinant DNA methods, immunofluorescence, high-content imaging, and live-cell fluorescence microscopy in human and mouse normal, cancer, and stem cell lines. Managerial:Ensured laboratory was in compliance with Environmental Health and Safety regulations. Ordered supplies and chemicals for the laboratory. Mentored and trained students, technicians, visiting faculty, and other postdocs.

Performed independent research and obtained reagents by contacting scientists all over the world. Skills used: Hypothesis testing, immunofluorescence and live-cell imaging, quantitative image analysis, Western blotting, antibody generation and purification, PCR, recombinant DNA methods, transient transfection, clonal selection, population statistics, immunoprecipitation, quantitative kinase assays, technical writing, public speaking. Research topic: CENP-A is phosphorylated by Aurora B (Zeitlin et al., Journal of Cell Biology, 2001)Detailed kinetics of H3 phosphorylation beginning in G2 (Zeitlin et al., Journal of Cell Science, 2001)Managerial:Ensured laboratory was in compliance with Environmental Health and Safety regulations. Ordered supplies and chemicals for the laboratory.