Work experience
Technical Fellow and Associate Director of ML.AI [Jan 2022 to present]
Principal Applied AI Scientist [July 2019 to Dec 2021]
Senior Research Scientist [July 2017 to July 2019]
Research Scientist [Feb 2014 to July 2017]
Led and managed the entire lifecycle of ML/AI digital health products from idea inception to FDA submission.
Fully designed, executed, validated and prepared FDA submissions for a deep learning based EEG record classifier aimed at reducing physician’s burden of reviewing patient data and improving patient’s health diagnostic assessments. Trained deep learning models with an unprecedented ~96% EEG record classification accuracy in new patients.
Built an AI based Therapy Recommendation Engine using contrastive and transfer learning techniques.
Recommended optimal device settings for treating new patients, by identifying previously treated similar patients using a kNN approach in the custom-learned EEG embedding space.
Discovered biomarkers of clinical outcomes in epilepsy by mining >3 million 6-minute EEG spectral images.
Used statistical and computer vision techniques for discovering robust EEG features predictive of clinical outcomes in large physiological datasets with noisy labels. Built and deployed ML models for classifying patients into outcome buckets.
Built EEG labeling tool, EEG search engine, and EEG record sorter.
Developed EEG spectral image embedding spaces with a range of deep learning models including deep ranking models, autoencoders, ResNets and custom CNNs. Clustered EEG records in the embedding spaces for enabling fast labeling of big datasets.
Characterized acute effects of brain responsive stimulation and effects of electrode implantation on brain activity.
Supervised interns and used statistical and ML methods for discovering stimulation-induced acute inhibitory effects. Discovered that brain activity varies substantially in the 1st 5 months after implantation of electrodes.
Actively led collaborations, publications and presentations.
Supported 20+ external collaborating institutions, wrote successful patents and peer-reviewed journal publications, and presented research findings internally and externally to a broad range of audiences, including board of directors.
Research Associate [May 2013 to Nov 2013]
Developed a closed-loop API for fast prototyping of algorithms for the brain stimulation platform, PC+S.
Built python and MATLAB tools that interfaced in real-time with the PC+S platform, and characterized the closed-loop latency of custom-developed brain stimulation algorithms.
Software engineering intern [July 2007 to May 2008]