• 2026/4 - present: Applied AI: Brewing Meta’s own “Claude Code”.

• Ads Training Data Infra Signal Growth

ML Feature Store

Building data platform and complex pipelines across multiple business domains (Zestimate, Zillow Offers & Mortgages) for both real-time and offline scenarios.

Tech Stack: Spark, Flink, Kafka, Airflow, Hive and S3

• Implemented the real-time-feedback machine learning platform (Multi Armed Bandit) at scale using Kafka Streams.

• Maintain and improve Expedia's main machine learning platform which productionalizes 40+ clients’ models (ex. hotel sorting), and handles thousands of requests per second.

Tech Stack: Java Spring Boot, Redis, Kafka

Cars Team (Back End)