Machine Learning Researcher
Current- Boosted the impact of Qure.ai's lung health products on over 4 million lives annually across 50+ countries by developing Computer Vision algorithms, which were independently confirmed to match or outperform.
- Principal architect of a scalable end to end training framework used to train over 100 models for 20+abnormalities on chest X-rays, streamlining collaboration in the team & reducing training time by 40%
- Enhanced data usability by processing over 50 million medical records via Extract, Transform, Load (ETL) pipelines, culminating in a curated trainset of 3.2 million and a testset of 300k entries
- Safeguarded business continuity by mapping critical data across 15+ devices (external hard drives & NAS) and backing up over 450 TB of data
- Streamlined the experiment tracking process and enhanced record-keeping by logging various metrics, image samples, and segmented outputs into ClearML, consequently establishing a reliable audit trail
- Developed a post-training module that enabled selecting, pruning, and exporting of model ensembles into TorchScript and ONNX, improving deployment flexibility across various runtime environments