Leading the Genomics Data Infrastructure team in the Data & Technology Engineering organization at Vertex.Group leader and principal architect of a flagship research data engineering platform integratingcustom built and SaaS tools for data capture, metadata annotation, pipeline development and deployment, workflow inspection, results publication, and integration with ELN systems. Dotmatics ELN templates allow scientists to export sample sheets and kick off sequencing and downstream analytics. AWS Event Bridge events from IDC are intercepted, data is registered in our custom database, copied to fast storage via Step Functions, Step Functions talk to our custom system where data and pipelines are registered and perform automated demultiplexing and execution of analytics pipelines based on stored parameter sets for each type of analysis. Execution is in Nextflow Tower. Results are published to Quilt Data, which allows easy "peeking at" and download of most data types. Analytics structured results and Quilt package URLs are published back to the Dotmatics ELN, completing the round trip. The entire data and process flow is visually inspectable and interrogatable via our REST API. Pipeline developers can make use of our cookie cutter repos and integrated (TeamCity) build system to automate pipeline deployments and dependency containers. Currently building a system atop DataBricks to enable researchers to use RStudio and Jupyter notebooks for downstream analytics.

Led a team building and managing Research Computing's Advanced Analytics Platforms.Architected and managed about 20 AWS developments for a variety of research projects, as well as classic HPC clusters at BCH Needham Datacenter and The Massachusetts Green HPC Center, running Slurm and Kubernetes. We provide an Open OnDemand portal to allow running ad-hoc jobs through a web interface, as well as cluster-hosted RStudio, JupyterLab, Matlab, and other programs. Developed a turn key collaborative data-sharing and analysis platform to support an ever-growing number of COVID-related studies, including the $30 million NIAID-funded COVID immunology project IMPACC: https://www.impaccstudy.org. A complete study website with AWS hosted domain, Cognito-based user administration, secured high-speed S3 bucket access, and integrated RStudio platform, and AWS Parallel Cluster using spot instances can now be deployed in one go, using a solution we built with Terraform, Ansible, GitLab CI/CD, and Vue.js. Built a big-data analytics platform to discover novel biomarkers in clinical time-series data, with a goal of automatically identifying and eventually predicting events requiring clinical intervention. At time of departure, the system was in beta in epilepsy studies and cardiac hemodynamics. An automated data pipeline pulls EEG and cardiac imaging data from clinical systems, extracts signals and metadata, performs various non-linear analyses using our Kubernetes cluster and Spark platform, and produces structured results and visualizations. We are in the process of abstracting and publishing this as a generic ETL workflow engine which meets the FAIR data principles for clinical research data.

Led a team producing a pre-clinical and clinical audiometry analytics platform.Developed a data lake and full-stack data analytics suite to process and analyze audiometry data, processing hundreds of studies per-month. Emphasis on automated big data pipelines, automated transformations and support for ad-hoc querying of large numeric datasets. Integrated common AWS services (EC2, S3, Aurora, SNS) with custom QC and analytics pipelines implemented in Nextflow, Pachyderm, and Python. Notebook-based UIs implemented in Apache Zeppelin and Jupyter/Papermill enable dashboard presentation and custom inline analytics. REST API and common data manipulations/export using Spark and Livy. Large-scale ad-hoc querying via Parquet-transformed datasets in Athena.

Technical lead on all Translational Science IT projects. Architected custom and vendor-supported systems on AWS and on-prem, to support biomarker discovery, target validation, preclinical efficacy studies, and preclinical imaging.AWS architect responsible for Translational Science systems. Architected many new scientific environments on AWS to support whole-exome, antibody NGS and microbiome analysis pipelines, ad-hoc data exploration and analysis, and proprietary tools. Implemented much of the AWS infrastructure personally, and managed the day-to-day R&D AWS support team.Technical lead on the integration of public and Takeda-proprietary NGS, real-world evidence, and clinical metadata into Takeda's R&D Data Lake. This included the creation of Hive tables and queries, migration of hierarchical data to Parquet, as well as writing experimental Spark pipelines for BAM and VCF file analysis. Technical lead on Takeda’s Scientific Knowledge Management platform, a multi-year, multimillion dollar project, which integrates data from file-share, Office 365, Documentum, Box, RDBMS, and AWS S3 into a common enterprise search system. Created hybrid AWS/on-prem architecture, managed implementation, and customized the ElasticSearch pipeline for Takeda proprietary data. Developed and maintained the Tumor Measurement System, which tracks treatment and measurement data for clinical pharmacology animal studies. Web-based interface to capture experiment details, animal dosing, and tumor measurements. Performs several types of statistical analysis, and generates IND filing reports. Instigated Takeda's transition to best-of-breed tools for infrastructure management (CloudFormation, Terraform, Ansible), version control (Git and GitFlow), and continuous integration tools (Bamboo). Administrator of the R&D DevOps and development tool suites.

Informatics Manager / Tech Lead, spearheaded the architecture and development of statistical analyses and visualization tools for expression analysis and translational research. Implemented an ahead-of-its-time web programming framework.