Internal:Created PoC data lake architecture, to package to clients using, spark, iceberg, glue, athena, and aws lake formation. Data governance would be provided via tag based access control from lake formation.Fintech Client:Led a team of 8 engineers that is at the forefront of setting up and maintaining a streaming data pipeline with 100+ sources and petabyte-scale data lake. This critical data pipeline and data lake allows the cyber ML/AI team to monitor threats.After evaluating productivity and velocity of the team, I implemented team reorganization following the Spotify model for team design. This resulted in increase of team velocity by 20%.Developed a strategic plan to transition from the Matillion platform to a serverless ETL platform, utilizing Snowflake Tasks. This move saved $125k annually and eliminated the need for infrastructure management required by the previous platform.Stack: Python, Snowflake, ELK (Elasticsearch, Logstash, Kibana), AWS (Fargate, ASG, Load Balancer, EC2, RDS, DynamoDB), Docker, Snowpipe, IAC (Ansible, Cloudformation), Alerting (Datadog, NewRelic), Kafka

Promoted after a year based on delivery and ability to mentor junior to mid level engineers across teams.Decreased time to deploy the ML/AI pipeline by 25% by moving manual processes to Infrastructure as Code and creating a testing suite to reduce errors.

Decreased AWS spend by 10% by creating a scalable serverless loading service that ingested data from our ML/AI pipeline into our application data store. Previously data was loaded in expensive and slow EMR jobs.Enabled our Client Ingestion Team to onboard and upgrade clients 15% faster by creating serverless automated tooling to run all components needed to build/upgrade our patient database.Stack: Kotlin, Java, JVM, Python, AWS (CDK, EMR, Step Functions, Lambda, ECS(Fargate), SNS, Kinesis, DynamoDB), Elasticsearch, Docker, GitHub Actions, Spark

Led the re-architecting of the data platform which increased the velocity of the team by 25%, and increased observability on all pipelines. This was done by using DataOps principles to create idempotent pipelines that could be developed/tested/deployed through the platform.Increased key game metrics for verifying healthcare providers by 10% by implementing an internal NPI API by using API Gateway/Lambda/Aurora (Postgresql). This process previously involved using contractor’s, and due to the manual process it would take weeks to verify vs seconds for the API.Stack: Snowflake, AWS(Lambda, Fargate, API Gateway, DynamoDB, Amazon Aurora - Postgresql), Prefect, Docker, Terraform, Serverless, Snowpipe, SQL, Python, dbt, Tableau

Led team of 4 that created our data analytics platform, with 120+ weekly users, that was instrumental in divvyDOSE being acquired by UnitedHealth. The platform was used for day-to-day operations prioritization & work allocation, product improvement, customer support, and integrated all necessary systems & components. The data platform (Snowflake, Looker, dbt, AWS) saw over 20x user growth.Guided team in creating idempotent data pipelines (streaming and batch) and data models (dbt) that provided valuable insights to the company across all departments by unlocking siloed data and combining previously unused/disparate datasets.Decreased Looker dashboard loading times by 70% by researching, planning, and executing a data warehouse migration from Redshift to Snowflake.Decreased data accuracy issues and data downtime by 20% by implementing data testing through dbt.Stack: Snowflake, AWS (Lambda, Kinesis, EventBridge, DynamoDB, SQS/SNS), Serverless, Snowpipe, SQL, Python, dbt, Looker, CircleCI, Stitch, ELT

Collaborated on projects with the Chief Data Architect and Chief Data Scientist at the State Data Practice. Created ETL pipelines for local governments to enable reporting and ingest data into the states open data portal framework. Created reports/dashboards in Power BI and Tableau. Stack: SQL, Python, Power BI, Tableau, Docker, Jupyter, MS SQL Server, PostgreSQL, MySQL, Pentaho ETL

Promoted to manage system analysts, program managers, and third-party vendors. Local Debt Recovery Program recovered over 150 million dollars for local governments and 13 million in program fees for the office.

Developed reporting methods/frameworks through SQL, Excel and VBA for Local Debt Recovery Program. Managed/Developed Access Database front end and back end for over 20 users.