Built out and led the Platform Team, tracking and addressing the systemic operational needs of Freckle: devops; CI/CD; A/B testing; data pipelining, analytics and visualization; curriculum development and publication; and leveraging the above for data-driven pedagogical evolution within the product.

We're leveraging machine learning and natural language processing to empower HR organizations to focus more on what matters most: people. AskHR is a platform for getting routine HR answers fast, and getting connected immediately to the right person whenever your needs go deeper than that. No more waiting for an email to get your policy number or update your 401k contribution. The result is threefold: happier employees, a streamlined HR department freed-up to devote more time to bigger issues, and robust analytic insight into the systemic needs of your organization.

I learned Haskell in a month and was lucky enough to be able to use this to empower millions of K-12 students across the US to learn Math, ELA, Science and Social Studies at their current education level. My time here involved building a system for nightly synchronization with diverse SIS systems, adding support for district-wide student benchmarks and analytics thereon, and improving a robust cyborgian pipeline for streamlining curriculum from ideas and drawings to YAML and images to database entities used by teachers and kids all across the country.

- Architected and delivered v2 of Trifacta's data profiling engine with improved emphasis on Apache Spark for distributed aggregate computation. Fixed our largest performance bottleneck and improved customer experience across the board- Architected and delivered integration with Apache Hive/HCatalog. Supports metadata exploration, project creation, and streaming ingestion/export for deep integration into existing customer data pipelines. Works in encrypted/Kerberized environments, opening the doors for our company to the entire existing Hive market- Architected and delivered integration with Amazon Redshift. Supports export of Trifacta results for downstream access to analysts in Redshift- Generalized support for Hive/Redshift into an open platform for JDBC connectivity. Enabled Oracle, SQLServer, Postgres and more to be treated as data sources within Trifacta

- Extended Trifacta's type system to support hierarchical map and array types. This paved the way for Avro/JSON/general schematized ingestion, restructuring, and export. Today, nested data types are table stakes for nearly all of our customers- Reworked Trifacta's data profiling engine to leverage approximation algorithms and abstract algebra for efficient, distributed numeric and categorical histogram calculation, anomaly detection, and statistical analysis- Designed and implemented Trifacta's first internal job graph model for script analysis, transform suggestions, and optimization

Worked on Espresso, LinkedIn's RESTful, real-time, NoSQL document store. Built a pipeline to/from HDFS for distributed bulk import and out-of-the-box ETL. Architected and implemented infrastructure to support live cluster expansion and rebalancing.

• Assisted with the development of Espresso, LinkedIn's horizontally-scalable, NoSQL distributed data store• Implemented Sphinx as an alternative secondary indexing search solution for Espresso, and used YCSB to benchmark its performance against Lucene• Investigated the possibility of real-time, distributed analytics under Espresso with the use of column-based storage engine• Configured a cluster of five machines to support up-to-the-second analytics queries on a test system with millions of rows being added every hour