Anmol Ranjan

Software Application Developer (a) @Google

Bengaluru, IN

Developed a graph-based ML system to detect fraud, utilizing semantic embeddings to identify \"fuzzy\" matches between abusive entities- Optimized graph generation latency, achieving full graph rebuilds in under 30 minutes for multi-million node datasets through efficient bucketing and sketching- Improved defensive recall by implementing a module that semantically links fraudulent business names previously missed by exact-match systems- Engineered high-performance data pipelines using Python and SQL to process large-scale datasets and construct similarity graphs- Developed LLM-powered agents for automated business verification, achieving an automation rate with high precision- Architected the agent using sandboxed browser environments and asynchronous messaging to handle high-concurrency requests- Integrated geospatial verification logic into the agentic flow, utilizing geocoding APIs to validate physical location data.