Software Engineer
CurrentSoftware Engineer at Microsoft on MAIDAP (Microsoft AI Development Acceleration Program)• Designed and implemented a C# API, enabling users to view accessible models, datasets, and associated runs at a per-sample level through Kusto queries. Enhanced user access to critical information and addressed a significant information gap. Developed Python infrastructure for automatically extracting model insights and developed an offline pipeline that reduced the necessary dataset samples by up to 140x smaller, while maintaining unbiased results. Offline pipeline was used to support runs in an AML pipeline. • Led the incubation of a self-healing feedback loop for customer feedback, collaborating with Applied Scientists to design an evaluation pipeline. architected and integrated a copilot-feedback solution using a llm-based evaluation using C#, Python, and GPT-4 through Azure. Integrated with four main Microsoft services, ensuring secure authentication and gaining buy-in from relevant stakeholders. Currently pursuing a patent for this innovation.• Scaled and migrated a Knowledge Mining API from Docker container (on physical machine framework) to an internal Kubernetes solution (on Azure Virtual Machine framework), impacting millions of tenant-based copilot users worldwide by speeding up how quickly they are served tenant specific knowledge.
