Mark Lewandowski Email & Phone Number

San Bruno, Ca, Us

San Bruno, Ca, Us

- TL/M for YouTube's Video Metadata team managing a team of 15 as we manage YouTube's internal Video data model and storage API, responsible for serving millions of QPS with sub 400ms response times.- Author multi-year roadmaps, working with cross functional stake holders to achieve alignment.- Responsible for hiring, mentoring, defining career trajectories, and performance evaluations.- Founding member of YouTube's "Core Entity" working group, where I work to develop principles and best practices for YouTube's core data model storage systems and APIs.- Migrated Video model from Python to C++ saving tens of thousands of CPUs and 10s of TB of memory in production. Awarded a company-wide, "Google Technical Impact Award."- Tier 2 oncaller, responsible for maintaining production excellence across 4 "critical" YouTube infrastructure services.

San Bruno, Ca, Us

- TL leading the design and implementation required to migrate YouTube from a 256-way sharded MySQL cluster (Vitess) to Spanner that guaranteed data correctness and zero downtime. Responsible for leading YouTube's Video Metadata team to move YouTube's canonical video model from MySQL to Spanner.- Owner of YouTube's canonical Video data model. Responsible for developing principles for working with and extending the model, and consulting clients to adhere to best practices.- Founding member of YouTube's Spanner schema council, responsible for developing and advocating for consistent schema design principles aligned with YouTube best practices.- Designed and built a novel notification system to send personalized recommendations to YouTube users.- Tier 2 oncaller providing 30 minute response times.

Edmonton, Alberta, Ca

- Provided key user analysis based on observed user behavioral patterns, and A/B test results (Redshift, Python, Pandas, Numpy, Scipy, Matplotlib, Statsmodels)- Managed development of internal and developer facing tools to simplify use of segmentation and communication APIs- Managed / primary developer on communication platform built using Python, Flask, Cassandra, Redis, Celery to provide access to platform developers to reach Betable users through multiple channels including email, SMS, and push notifications- Managed funnel optimization through experiments utilizing A/B tests based on observed patterns in data, increasing funnel conversion by 15% over 6 months (Javascript, Node.js, Mixpanel)- Managed segmentation API enabling internal marketing and external developers an interface to segment users into cohorts matching query criteria using Javascript, Node.js, Cassandra, MySQL, and Redshift (Postgres)- Responsible for providing best practices and performance data to external developers through regular communications, and presentations- Managed / co-developed ETL infrastructure to move data from Cassandra through Hadoop (via Pig) to Redshift, enabling a more robust analytics infrastructure

New York, Ny, Us

* R&D Lead for Digg Ad Server, a highly scalable project designed to serve Digg based advertisements on Digg.com and syndicated across the web. Researched and developed first, second, uniform, and vickrey clarke groves auction algorithms. Built extensible targeting system for taxonomy, keyword, geographic, and user based ad targeting. Developed dynamic models for impression forecasting, and click-through rates.* R&D Lead responsible for building a new duplicate content detection algorithm using Locality Sensitive Hashing (Minhash/Simhash) to quickly detect duplicate/related content already contained within the Digg corpus in sublinear time.* Researched and developed algorithms to find trending topics and their related subtopics from sources across the web to identify breaking news.* R&D Lead for researching and developing a content based recommendation algorithm now featured on Digg.com story pages. * Designed, implemented and deployed services and command line tools for site developers to use technologies grown in R&D (Python, REST, WSGI, Apache, MySQL, NoSQL)

Livermore, Ca, Us

* Designed, implemented and performed performance analysis on particlelocation algorithms to determine which would scale the best for petascalecomputing on machines including Blue Gene/L and ASC Purple.* Refactored data structures for reuse across different particle locationalgorithms using C.* Automated major test cases using Python.* Developed rules to enforce safe coding practices in C and C++ throughtraversal of a source to source compiler's abstract syntax tree. The ruleswere implemented in C++. Project is now opensourced (http://rosecompiler.org/)* Included documentation for each code checker to be included in release toopen-source community.* Helped develop and implement code certification process.

Tallahassee, Fl, Us

Tallahassee, Fl, Us

Phd (Abd), Computer Science

Ms, Computer Science

Bs, Computer Science