AI Engineer
Company Context
Microsoft Research — the research arm of Microsoft, Cambridge UK lab. Worked under the direct supervision of Pushmeet Kohli (now VP of Science & Strategic Initiatives at Google DeepMind), supporting experimental ML research. Contract engagement, February to May 2011.
Summary
Supported experimental ML research by building data-driven web and mobile applications that bridged academic models and scalable real-world tooling. Focused on human-in-the-loop data collection, MLOps-style pipelines, and internal research tooling for Kinect, Photosynth, and early facial/body tracking models.
Responsibilities
- Developed Facebook-integrated web and mobile apps to crowdsource labelled data (images, video, text) for training machine learning models — an early form of human-in-the-loop data collection
- Designed and implemented MLOps-style pipelines using Windows Azure Cloud Services, handling ingestion, validation, storage, and asynchronous processing at scale
- Created internal tools to support dataset visualisation and annotation workflows for ML researchers working on Kinect, Photosynth, and early facial/body tracking models
- Delivered Silverlight apps using MVVM architecture on Windows Phone 7 to capture sensor-driven input from mobile users, later used to refine model assumptions
- Built backend APIs and dashboards using ASP.NET MVC 3, Razor, and SQL Azure, with secure user authentication and Facebook graph integrations
Outcomes
- Enabled crowdsourced data collection at scale for ML model training
- Delivered MLOps-style cloud pipelines for ingestion, validation, and processing
- Provided internal research tooling supporting Kinect, Photosynth, and body tracking research
Reusable CV Bullets
- Developed Facebook-integrated web and mobile apps to crowdsource labelled data for training ML models — an early form of human-in-the-loop data collection
- Designed MLOps-style pipelines using Azure Cloud Services, handling ingestion, validation, storage, and asynchronous processing at scale
- Created internal tools supporting dataset visualisation and annotation for researchers working on Kinect, Photosynth, and early facial/body tracking models
- Delivered Silverlight apps using MVVM on Windows Phone 7 to capture sensor-driven input for refining model assumptions
- Built backend APIs and dashboards using ASP.NET MVC 3, Razor, and SQL Azure with Facebook graph integrations
- Worked under the supervision of Pushmeet Kohli (now VP at Google DeepMind), bridging academic ML research and scalable real-world tooling
Evidence / Source Notes
- Source: verbose CV enrichment; reconciled with JobVia export (
madu_alikor_export.json). - Confidence: medium — detailed role description from CV
Open Questions
- Volume of labelled data collected through crowdsourcing apps
- Specific ML models or papers supported by the tooling