Lead team of 3 SDE to develop a training data pipeline to ingest and federate 1500 feature data across IG and Oculus organization for content serving. Automatically validate, preprocess, and optimize access to prediction models. Optimized feature-selection and ranking algorithm. Improved truth label and regression accuracy from 0.8975 to 0.9124. Proposed and developed a new asynchronous pre-fetch pipeline to load customer profile and behavior signal, and a multi-level cache mechanism. Reduce critical personalization UX loading latency from 2.5s to 0.5s. Defined canonical data schema to standardize access for both near-real-time event data and batched ETL ingestion. Developed an observability platform to monitor pipeline health and congestion in real time with extensible analytics. Developed an GenAI-based advisor agent for context-aware action, and knowledge-base with unstructured SOP data. Pioneered GenAI tools for developer productivity improvements, evangelized vibe coding, review and design standards. Developed the complete Global Message Payment full-stack application which facilitates seller transactions. Planned and executed cross-functional product requirement analysis to scope and plan for end-to-end project execution. Architected a set of micro-services for complex payment interaction, such as payment slip management service, app-switch flow for third party payment integration, and sensitive information retrieval compliance enforcement mechanisms. Developed an GenAI-enabled onboarding automation for payment gateway integration and configuration, automatically generating artifacts. Eliminated human modeling and custom coding work for adapters and integrations. Successfully launched applications and developed an extensible architecture which now scales to 50+ countries.
