At Microsoft, I architect and optimize AI-driven Data Enrichment APIs, refining system prompts and implementing resilient post-processing to improve data standardization accuracy and product reliability.
I also design cost-optimized next-generation data platform architectures, enhance observability for faster triage, and deliver customer-facing product documentation. Previously, I developed Azure Hyper-Converged Infrastructure features and onboarded the “Deploy from Cloud” flow to the disconnected Azure Arc environment, reducing cluster creation time by 95%.
Earlier at Microsoft, I created a proof of concept for a Microsoft SharePoint video playlist gallery view. In my projects, I implemented a Transformer-based handwriting recognition model that achieved a 7% word error rate, and implemented research to detect atomicity violations in a NodeJS application.

