Sonu George Email & Phone Number

Boston, Ma, Us

San Francisco, California, Us

- Led a team of 4 engineers to design and implement new pageview datasets in Snowflake Data Warehouse, utilizing multiple batch data pipelines to process 2 million daily impressions with newly revamped front end logging that better captured the user's interaction with Nerdwallet reducing duplicate page views by 25%- Spearheaded the decommissioning of the old data warehouse system sunsetting ~2K critical tables as part of a multi phase project spanning 6 months resulting in substantial savings of several hundred thousand dollars- Architected the Near Real Time enrichment pipeline using Kinesis and Lambda that enabled taxonomy enriched data to be available to all end users almost instantaneously replacing the manual process that took multiple analyst hours each week which in turn improved data quality by eliminating manual entry errors- Designed the first MVP for front end logging using Segment event instrumentation payload working closely with Engineering, Analytics, and Product teams to showcase the end to end flow of data in the new system whose success led to the launch of the companywide project to revamp front end logging- Created preprod environment that replicated the end to end flow of data from source tables all the way to the aggregate dbt layer including corporate dashboards that enabled us to validate big pipeline changes and backfills seamlessly ensuring quicker sign off from Analytics and leadership team when critical metrics were involved- Developed an automated system to efficiently process CCPA data requests identifying and anonymizing user's corresponding event data replacing the manual process that took over two hours per request Technologies: Python, Airflow, dbt, Snowflake, SQL, Segment, Kinesis, Lambda, APIs

San Francisco, California, Us

- Created the end to end data pipelines needed to ingest event-level data into the companywide self service platform enabling all employees to have easy access to crucial business data- Designed and developed the data architecture for offline data ingestion to power Nerdwallet's AI/ML models consisting of similar product recommendations, product rankings, and user classification ML models- Designed the feature store prototype which enabled the Data Science team to train models faster by having the data in the same location as the ML models- Won the company-wide Best Show Hackathon Award for the week long crypto hackathon project integrating crypto currency wallets and exchanges into NerdWallet leveraging APIs for real-time data synchronization Technologies: Python, Snowflake, SQL, Airflow, S3, SageMaker, Amplitude, Redshift

Menlo Park, Ca, Us

- Led pilot program to enrich shopping data in realtime via API calls, mobile logging, and object data models as part of companywide initiative to ensure better metrics across systems- Architected and built the end to end data pipelines in Spark for the launch of Shopping Drops product which captured the overall user journey from the initial browsing all the way to when they made the purchase providing detailed insights on both user attribution and revenue attribution- Collaborated with front-end engineering to design and implement the logging roadmap for the Creator project providing fresh insights into the daily activities of Influencers in marketing products directly to consumers- Created the time spent visualization dashboard allowing top leadership to quickly answer high level questions on how users are spending their time on the app- Launched temp tables bot that reduced petabytes of Hive data warehouse storage by automatically cleaning unused tables saving thousands of dollars in warehouse costs on a weekly basisTechnologies: Python, Airflow, Presto, Spark, Hive, Hadoop, SQL

Scottsdale, Arizona, Us

- Revamped failing data mart by working hand in hand with data architects and database developers to launch a new data warehouse that integrated wide range of brand new data sources- Designed an online vendor recommendation system that ranked 35k vendors using multiple KPIs and metrics to identify the best matched vendor to specific service calls reducing incorrect assignments by 80% boosting profitability- Created the performance metrics scorecard for 14 major operational roles that enabled the executive team to directly tie employee performance to organizational goals resulting in 75% quarter over quarter improvement in employees meeting their objectives- Analyzed revenue invoicing data across multiple systems to identify and understand the various stages of the billing process leading to actionable insights and significant process improvements companywideTechnologies: Python, Web Scraping, APIs, MySQL, Oracle, SQL

Master Of Business Administration (Mba), Management Information Systems

Bachelor Of Engineering, Computer Science