Jul 2025 – Present · 10 mos · Hyderabad · On-site

• Working in the Maps org on the Geo Data Platform team. My day to day is Spark/Scala pipelines that handle geospatial data ingestion, conflation, and validation for Apple Maps across 197 countries.
• Some things I've worked on:
• Optimized a probe data ETL pipeline (~100 TiB/month across 10 countries). Brought runtime down from 10+ hours to 4.6 hours and cut monthly costs by 70% through serialized persistence, S3 checkpointing, and a tiered executor config (200/500/1000 executors based on country volume).
• Found and fixed a bounding box bug in building conflation that was creating 10.3 million false deltas per cycle. One line fix, 97% reduction in noise.
• Fixed a turn restriction bug that was incorrectly blocking valid turns on highways in 3 countries. Traced through 500+ lines of Spark transformations to find the root cause in path selection logic.
• Built a Spark pipeline to refresh a stale airport geometry resource (untouched since 2019) by extracting from 1.1B+ OSM ways. Expanded coverage from 15.6K to 23K airports across 238 countries.
• Debugged and stabilized 3P address ingestion pipelines across multiple countries during EMR to EKS migration. Fixed dynamic allocation issues, S3 URI scheme conflicts, serialization bugs.
• Extended the OSM validation framework with 10+ data quality checks and fixed MapRoulette upload logic for per-country projects.
• Tech: Spark, Scala, Java, AWS (EMR, EKS, S3), Parquet, Protobuf, Kafka, Redis, Cassandra, Kubernetes