At L&T Finance, I integrate gold-loan data into the Enterprise Data Warehouse, delivering DIMENSION and FACT models for cross-sell and upsell analytics. I also optimized existing loads, reducing end-to-end runtime by 70–80%.
I optimized Vertex AI credit-risk batch workflows covering 200K+ customers, cutting processing runtime by 80%. My work also includes Loan Origination Platform data integration and maintaining data integrity across high-volume bureau and core-banking records.
At DeNA Co., Ltd., I developed a user-engagement classification pipeline processing 80K+ users per run with 99.5% classification success. I also engineered an AI-powered thumbnail authenticity detection system and built projects in particulate-matter prediction, multi-PDF question answering, and semantic video search.
