I lead the Applied ML team at Scale AI, where we build ML and LLM systems that improve the quality and efficiency of human annotation pipelines powering models like ChatGPT, Gemini, and Llama. My team tackles challenging problems: validating expert-level scientific and reasoning data, maintaining annotation accuracy at scale, building human-in-the-loop ML tools that multiply annotator effectiveness, and creating infrastructure for rapid experimentation and deployment.
Before Scale, I led a Copilot team at Microsoft AI that trained and deployed multimodal LLMs optimized to run on Windows devices without cloud dependencies. This work was central to the Copilot+PC initiative. I also led Microsoft's AI efforts in life sciences, where my team used ML to optimize clinical trial designs and accelerate drug discovery, partnering with Fortune 100 companies to extend these techniques to discover novel chemical compounds for sustainable energy technologies.
Earlier, I worked at Facebook AI Research (FAIR) in Paris, Siemens in Princeton, and Amazon Research in Cambridge. My research has been published at top-tier venues including CVPR, ICML, WACV, and ICCV.
Experience
Head of Applied ML
Scale AILeading the Applied ML effort at Scale AI. We build ML and LLM tools that make human annotators more effective at labeling data for models like ChatGPT, Gemini, and Llama. The core problems we tackle: using ML systems combined with human feedback to validate complex scientific and reasoning datasets, and creating multi-agent workflows that catch errors and speed up annotation pipelines.
Principal Research Manager
MicrosoftOur team works on the following domains: LLM training and fine-tuning, compression and quantization, LLM evaluation and responsible AI, Tool/Agential training, and multi-modal integration. Our work contributes to the Copilot+PC feature.
Applied Science Lead
MicrosoftLed a team of Applied scientists to build and deploy ML solutions to optimize clinical trial design and accelerate drug discovery. Responsible for setting the technical direction for the team and communicating the plan to customers. Individually contributed and led the development/deployment of large-scale projects involving LLMs, entity extraction, knowledge graphs, and image generation.
Applied Scientist 2
MicrosoftDeveloped ML solutions for Healthcare and Clean Energy domains from idea conception to product pipeline deployment. Worked on developing Computer Vision and Natural Language Processing state-of-the-art solutions for production scale problems.
Applied Science Intern
AmazonMetric transfer learning for Dash Cart Project.
Research Contractor
Facebook AI Research (FAIR)- Diverse Generations using Determinantal Point Process (GDPP) (ICML-19: Spotlight)
- Zero-Shot Learning via Creative Adversarial Learning (ICCV-19: Poster)
- Multi-modal future predictions upon a stochastic prior (CVPR-21)
Research Intern
Siemens- Generative Adversarial model to synthesize unseen views of natural images (Patent)
- Multi-view egocentric video summarization (WACV-20)
Education
Doctor of Philosophy - PhD
University of Central Florida
Computer Vision & Machine Learning
August 2016 - December 2020Bachelor's degree of Engineering
Alexandria University
Computer and Systems Engineering
2011 - 2016