My Work
Currently working on solving ad-tech problems at scale. I code for passion & design for fun. On a mission to build efficient and accessible products and help the next generation of devs along the way.
Career Journey
Where I've Worked
Pubx AI is an AI-driven price floor optimization platform that helps publishers increase ad revenues through dynamic floor pricing, targeting a minimum 3X ROI.
Key Contributions
- Built a comprehensive automated E2E testing and validation pipeline using Playwright and LambdaTest that validates client tags across live publisher sites in multiple geolocations and browsers. The system processes test results through Kafka queues, stores data in Parquet format on S3, and powers both automated alerting and Grafana analytics dashboards. This pipeline reduced manual QA time by 50% while proactively catching production issues.
- Led the complete rebuild of the publisher-facing dashboard from ground up, collaborating closely with product and UX teams. Implemented the core architecture using React, TypeScript, and shadcn/ui components based on Figma designs, along with a Django backend API. The new dashboard improved publisher engagement by 40% and reduced support tickets by 60%.
- Optimized data aggregation pipelines for Classification and DBM models by migrating from AWS Kinesis to managed Kafka (WarpStream), reducing monthly AWS costs by almost 45%. Built a Python-based data validation service using Pydantic that ensures data integrity before ML pipeline consumption.
- Currently developing Stream, Pubx's new Tag Management & Automation Service that captures additional datapoints from JavaScript tags and combines them with realtime auction data consumed by decision tree algorithms. This has resulted in an 80% reduction in manual configuration overhead.
Started my journey in 2019...