Sina Serati

Stack: Python, SQL, Databricks (Delta Lake, Jobs, Dashboards), Angular, Azure, AI/LLM tooling

• Moved financial, AI, and procurement dashboards out of manually maintained Excel workbooks and into Databricks, so finance, procurement, and leadership teams can pull current numbers themselves
• Built custom Databricks tables on Delta Lake with SQL and Python, rebuilding the spreadsheet logic as reusable datasets and metrics behind the dashboards
• Worked with the data team to build and deploy scheduled Databricks Jobs that refresh the tables and dashboards automatically, replacing manual spreadsheet updates
• Shipped the new AI-IRB application on Azure: users upload a study protocol and the app fills out the review board’s full questionnaire, cutting protocol intake from days to minutes
• Build features and fix bugs on WCG’s IRB (Institutional Review Board) platform, a set of Python services with an Angular front end, including board reporting for 5,000+ protocols per year

Stack: Python, Django, Celery, Pandas, PostgreSQL, RabbitMQ, Checkout.com, AWS (Lambda, SQS, DynamoDB, S3), Docker, Datadog

• Lead engineer on a new event-sourced billing and ledger system, taken from design to production in 3 months. It runs scheduling, cancellations, and payments for $54M/month across 5,000–10,000 payment events/day, with full traceability, replay, and upcasting for backward-compatible schema changes
• Built mid-term adjustment (MTA) price change flows that recompute billing schedules for 300–1,000 of 400k active policies each day, plus prorated refund and cancellation engines (first purchases run synchronously; refunds and follow-up actions are event-driven)
• Built ACH return handling and check payment flows with configurable recovery actions and Celery Beat dunning schedules, raising payment recovery rates by 80%
• Built multi-gateway payment orchestration with a volume-tier pricing engine ($5M and $15M thresholds) and worked on PayFac implementations covering $100M+ in transactions, generating $1.7M in new revenue
• Added Datadog tracing across async Celery task chains so failures alert immediately instead of going unnoticed; automated corrections and event backfills recovered hundreds of thousands of dollars
• Shipped a self-service rating engine that cut new-state launches from months to one day, used across all 50 states

Stack: Python, Flask, Scala, ReactJS, AWS (Lambda, API Gateway, DynamoDB, SQS, Step Functions, Glue), OpenShift, Kubernetes, GitLab CI/CD, Splunk, Terraform, HBase, Hive

• Built a serverless data platform on AWS Lambda, API Gateway, and DynamoDB with OAuth/ISAM authorization, VPC isolation, and role-based access control
• Designed a CDC ingestion pipeline streaming millions of records/day from on-premises databases into an AWS data lake, with Spark jobs orchestrated by Step Functions and monitored in Splunk
• Contributed to a Data Mesh architecture and a rule-based IAM policy system that removed data duplication and let data scientists and actuaries query cloud data directly, which opened up new lines of business