Leading and building a seamless, secure, and scalable event logging platform to empower data-driven innovation and decisions across Airbnb.

Technical lead focused on increasing developer velocity and improving developer experience for Go application development. Built reference application framework and drove alignment and adoption across engineering.

Bootstrapped and led rideshare expansion into multiple international markets. Managed overall program across 20+ cross-functional teams. Distilled technical architecture, dependencies, risks, and sizing for senior leaders and executives. Led technical discussions with 3rd party vendors and demonstrated regulatory compliance with city regulators.Led data infrastructure and quality improvements to accelerate model training, testing, and serving of Lyft’s forecasting of rideshare market outcomes to further revenue and growth goals.

Advised Executive Leadership Team on all technical aspects of the business, including engineering team structure and processes, software architecture and stack, and roadmap and prioritization. Designed and led development of core data pipelines to power flagship products. Evangelized Change-Data-Capture (CDC) and laid the technical foundation for enabling state synchronization across real-time reactive systems.

I am a founding engineer of Lyft’s Streaming Platform team. Helped build the team from four to twenty-five engineers and the overall technical lead across the platform. Teams include PubSub (Kinesis/Kafka), Ingest & Persistence (streaming/enriching/persisting all analytic events), and Data Quality (improving the discovery, understanding, and overall quality of online and offline data). Integrated Apache Flink into the Lyft stack making it a first-class citizen for engineers to more easily write streaming services. Worked closely with several data customers to debug and optimize their Flink jobs. Implemented multiple Flink jobs and connectors for stream processing and storage.Re-architected Lyft’s analytic event pipeline. Reduced time from event logged to user query from 36 hours to a game-changing p99 of 5 minutes. Reduced costs by $4MM/yr. Consolidated to a single wire format and enforced syntactic and semantic validation at the edge. Implemented a Flink job to perform Parquet translation and stream to S3. Extended Flink to provide a fault-tolerant, global view of application state to enable use cases like event-time based success files for orchestration of downstream ETL.Advised executive leadership with a multi-year roadmap and built up a team for moving Lyft’s messaging infrastructure from Kinesis to Kafka to more reliably and cost-effectively support business use cases.Built a Presto-based assertion framework for continuously monitoring offline data for semantic correctness. Re-architected Lyft's data pipeline completeness monitoring system to be more scalable and reliable.Led cost reduction efforts across the company with a focus on optimizing core streaming services. Worked closely with customers to profile their service and remove bottlenecks. Reduced costs by $6MM/yr.

Tech lead for the Ads Spend pipeline. Maximized ad revenue using a lambda architecture to calculate live spend based on real-time impressions and engagements compared against campaign budgets.Tech lead for the Insights Audience API, which supports interactive query of aggregate statistics describing the demographic characteristics of custom groups of Twitter users, in a privacy-focused environment. Applied probabilistic data structures (Hyperloglog, Locality Sensitive Hashing, MinHash) to minimize operating costs and query latency and techniques in differential privacy to ensure user privacy.Productized the delivery of high-fidelity impressions to global partners based on a set of tweets. Collaborated with other engineers designing, developing, and testing the Scala RESTful API and backing services. Implemented blackbox external tests that run daily and alert stakeholders when discrepancies and failures occur.Contributed an annotation-based validation module to the Finatra web framework. Field- and method-level annotations supported. Implemented several custom validators to support customer use cases.Implemented a proof-of-concept workflow to deliver daily tweet authorship and viewership demographics (age, gender, region) and statistics. Workflow powered Twitter's election website for the 2014 US elections.Refactored a monolithic, internal client testing framework into a layered web application with a RESTful API. Added support for data persistence using finagle-mysql and MySQL. Mentored team in REST principles, relational databases, and integration testing.

Designed and implemented clustering software for horizontally scaling Apache Nifi. Features include dynamic application discovery by way of multicast, high-availability of a centralized cluster manager, centralized security, and configurable secure communication using HTTPS and secure sockets. Largest cluster size in production is over 1000 nodes.Implemented a HTML5 web application enabling users to create hypervideos by enriching existing videos. Enrichments include time cropping, embedding hyperlinks, images, and commentary, and censoring frames via blackouts.Implemented an external Domain Specific Language (DSL) enabling developers, sustainment engineers, and users to more easily query a proprietary, binary file format. DSL queries on average required 8X less memory and 3X less CPU than alternative strategies.Led small teams (3-5ppl) to modernize several operational data flows. Designed and developed multi-threaded data flow processors for the NiagaraFiles framework. Processors unit tested using Mockito.

Implemented Row Level Access Control in Sybase to securely manage data access to an existing knowledge base. Implemented a software security layer that authenticates users using either username/password or a PKI certificate. Used jQuery to develop a flexible query workflow enabling users to build queries dynamically through a web interface.Co-Led an internal research and development effort focused on extending the Graphical Analysis Report (GAR) specification to support Multi-Intelligence metadata. Engineered a MySQL database schema for supporting dynamic metadata definitions. Implemented an XML-based service API to submit, retrieve, and query GAR metadata. Demonstrated system capabilities to potential customers resulting in additional funding and new customer relationships for the company.Designed and implemented a domain model and business processes using Java and Spring for a subsystem of the DCAS OpenCable Application Platform. Unit tested components using JUnit and EasyMock. Assisted in developing a Hibernate-based persistence layer. Implemented a CD Burner application by wrapping the cdrecord UNIX utility in Java.Significantly contributed to large-scale proposal effort including RFP review, requirements analysis, system design, and technical writing. Wrote 25% of the technical volume.Prototyped a cross-platform Ajax application for query federation and asynchronous result set display. Evaluated several open source Ajax toolkits (Dojo, Yahoo User Interface, Prototype) for streamlined web development.

Led a small team (4ppl) of engineers in implementing and integrating intelligent processes for Electronic Intelligence (ELINT) processing. Implemented algorithms include a fuzzy inference system correlating signals to sites and a scalable order of battle comparison allowing analysts to efficiently determine discrepancies.Provided team members with training in machine learning techniques strengthening their technical background and improving team communication.

Applied techniques in artificial intelligence to SIGINT.