Mark Schatzman Email & Phone Number

Stockholm, Stockholm County, Se

Developed key analytical datasets that supported Spotify's strategic initiatives to reach 29.2% gross margin.Enhanced and optimized Spotify's most widely used analytical dataset, improving data accessibility and insights for stakeholders.Achieved 25% reduction in costs and run-times across datasets and pipelines by implementing innovative join optimizations and compression strategies, demonstrating a strategic approach to enhancing efficiency and performance.

Stockholm, Stockholm County, Se

Spearheaded the implementation of sort merge bucket (SMB) optimizations for one of the most widely used datasets at Spotify, mitigating Dataflow shuffle costs, decreasing overall runtime by 15%, and enabling more efficient joins for hundreds of downstream consumers.Developed Scio pipeline to process 100+ TB's of licensor data through utilization of monoids and aggregators. Empowered gross profit predictions.Drove refactoring of core Python and Scala abstractions through inheritance and mix-in functionality to DRY up code, resulting in a 25% line reduction and increasing test coverage to 96%.Instructed DE 201: Introduction to Data Engineering.Taught programming languages, software architecture, and pipeline development to aspiring engineers in the 2022 NYC Tech Fellowship program.Designed and implemented offline click through rate (CTR) feature construction through Scio batch pipeline, enabling a ranking model to be trained at the uri-country and uri-query granularity.

Stockholm, Stockholm County, Se

Empowered hundreds of Finance users through the creation of a data layer that enabled granular insights through observations of actuals and forecast data.Elevated the functionality of Scio pipelines by introducing real-time reconciliation checks and enabling early identification of anomalies, thus substantially enhancing data processing integrity.Allowed FPA users to configure mapping changes for the functional PnL, through the design and creation of a pipeline which leveraged a recursive algorithm to parse and detect updates on a daily cadence.

New York, Ny, Us

Cut costs by 30% month over month by fortifying data through the reduction of DAG failures, prevention of cascading issues, and enhancement of efficiency for Spark and Hadoop jobs.Established feature store in Postgres, optimizing data pipelines and introducing services for streamlined ML development.Driving force in enhancing analytics through the development of ETL pipelines in PySpark to move data between Postgres/MySql and Snowflake.

London, England, Gb

Increased revenue 3% by enabling deeper insights into audience profiles through contributions to Spark pipelines.Helped build out user id space by creating features for a Java service that improved our match rate with DMPs. Proactively spearheaded the effort to update applications—Introduced more resilient third party libraries and optimally designed algorithms. Led effort to migrate from antiquated hourly Luigi orchestrated Hadoop Map-Reduce jobs to Lambda triggered Airflow orchestrated Spark jobs. This detangled job dependencies, and pronounced the structure of data workflows, causing a discernible reduction of time spent fixing broken workflows. Created Spark pipeline to process TBs of data. This workflow processes campaign performance data, user ids and audience profiles; additionally supports inventory quality compliance and high performing implementation of data science algorithms. Resulted in improved KPI performance for activation on the platform.

London, England, Gb

Enhanced bid request fraud detection by 20% through creation of daily orchestrated Spark pipeline to process ads.txt data.Created TCP/IP pixel server, empowered by Netty, focused on high throughput and low latency. Alpha version performed up to 10x better than the original Node.js server during benchmark RPS testing.Engineering lead for Java pixel server using Netty. Orchestrated deployment through Docker and ECS.Monitored data ingestion and manipulation workflows, resolving issues as they arose by deploying appropriate modifications to application code and server configurations.

New York, Ny, Us

Analyzed business presentations in teams and presented feedback and recommendations to various executives.Utilized Amazon Kinesis and many other external APIs to achieve a scalable and durable NIO server with clean, thorough analytics.

Allentown, Pennsylvania, Us

Improved legacy features for new Tracer Engine.Developed RESTful web services using Microsoft’s WebAPI.