Sep 2025 – Present · 6 mos

• As VP for Machine Learning for Office AI, I lead the vision, strategy and development of AI for Copilots within Microsoft 365.
• End to end Leader for horizontal capabilities (Podcast, File Comprehension, Evals) for Office AI
• Strategic Guidance for leveraging New Industry leading models (Open AI, Anthropic, OSS).
• Fine-tune models with Gigascale throughput and industry leading latency and COGs, using SFT, Reinforcement learning.

May 2024 – Sep 2025 · 1 yr 4 mos

• As the Partner Group Science Manager at Office AI, I lead the development of AI for Copilots within Microsoft 365. My team is responsible for various Office Copilot skills like Document Q&A, PowerPoint summarization. We build common ML infrastructure for large scale data mining, evaluation. We collaborate with all Office and broader MS teams developing Copilot experiences. Finally we build fine-tune models which have significant lower latency and COGs, experiment and innovate scenarios as new model families and capabilities arise.

Mar 2021 – Apr 2024 · 3 yrs 1 mo

• Managing a global team (Seattle, EU, India) with end-to-end delivery responsibility for Microsoft Machine Translation. Responsible for all parts of the ML pipeline, data acquisition, model building, deployment to the cloud (Kubernetes architecture), live site (Azure Translation API), and the Translation business for Microsoft.
• The Azure translation API is used by thousands of developers to build innovative solutions. We deliver translation experiences that serve hundreds of millions of users worldwide.
• Leveraging LLMs to enhance translation scenarios and improve LLMs for multilingual scenarios.

Oct 2017 – Feb 2021 · 3 yrs 4 mos

• Collaborated with teams across the company to unify the localization stack for Microsoft, saving the company millions annually.
• Launched various products from ideation to V1 with both engineering and business ownership, including Document Translation, Custom Translation, and Group Transcribe.
• Experienced in hiring and building teams across the stack, including Mobile Apps, Distributed Systems, AI, and NLP.

Jan 2014 – Sep 2017 · 3 yrs 8 mos

• Development Lead for https://hub.microsofttranslator.com/ and Backend Runtime service @ http://www.bing.com/translator for Microsoft Translator.
• Continuous bi-weekly delivery of features, performance and model improvements.
• I love being hands-on and writing code. Spent ~ 50% of my time writing code, debugging and solving various customer asks and live site issues.
• Built expertise in large scale ML systems, NLP, building and supporting high scale cloud services and now Deep learning. Completed Udacity DNN Nanodegree.
• Paper describing part of the work -- EMT: End To End Model Training for MSR Machine Translation
• http://dl.acm.org/citation.cfm?id=3076247

Nov 2010 – Jan 2014 · 3 yrs 2 mos

• When I joined the team, I brought in service experience learnt at Azure and picked up Machine Learning + NLP chops needed for Machine Translation. Took various online classes @ Stanford, Coursera for in-depth subject understanding.
• Own the data acquisition pipeline and model building for Machine Translation. The data pipeline collects the training data, through both well published as well as the entire web index, and transforms and filters it to be used for training data for building the translation models. Training infrastructure entails running Hadoop + HPC jobs to build the models that eventually serve translation requests at runtime.
• Built scalable transliterator training & runtime service.
• Designed crowd-sourcing infrastructure to close the loop from our log data to improved quality.
• Presented at STRATA & GTAC.

Aug 2004 – Nov 2010 · 6 yrs 3 mos

• o Successfully delivered various milestones & V1 release of Azure Service Bus.
• o Owned Monitoring feature in Dublin & NetFx 4.0 WCF/WF.
• o Biztalk Server tracking and messaging runtime.