Senior Data Scientist - Team Lead
2019 - PRESENTNielsen- Built predictive models to provide useful insight into the clients' product.
- Set up Airflow to automate ETL workflow.
- Implemented a messaging queue to manage model training jobs.
Technologies: PythonData Scientist
2017 - 2019APN Health- Built object detection models for real-time tracking of catheters, patches, and cryoballons in x-ray images.
- Built deep learning-based model for classifying heart arrhythmia from ECG (heart) signals.
- Developed a 3D Convolutional Autoencoder for generating realistic 3D heart chamber map from very few 3D points.
- Built dataset annotation system. Developed both client and server-side applications.
- Developed RESTful web service endpoints for object detection and ECG classification models.
Technologies: C++, C#, Java, Azure Cognitive Services, Keras, TensorFlow, PythonMachine Learning Engineer
2016 - 2017GE Healthcare- Built GE Healthcare Tube Watch system – Tube Watch is GE Healthcare’s predictive solution that is designed to remotely monitor tubes and predict failures before any disruption occurs.
- Designed and implemented multi-modal failure prediction algorithms for CT machines.
- Developed an NLP-based classifier to identify failed parts in CT machines using field support data.
- Developed optimized data preprocessing system that collects and converts unstructured big data and into structured database records. Processing about 100 million records in 2 hours on a single machine.
Technologies: PostgreSQL, MongoDB, Bash, Java, Node.js, MATLAB, PythonData Analyst
2014 - 2016GasDay- Developed a natural gas demand model-adjustment algorithm that accounts for behavioral impact on gas demand.
- Developed a pattern recognition algorithm to identify certain rare events in gas time series data.
- Developed daily gas demand forecasting models to help gas utilities in their daily operations planning.
Technologies: MATLABResearch Engineer
2012 - 2014Bioinstrumentation and Neuroengineering Lab- Implemented a 1-D cursor control with EEG signals using FFT and logistic regression.
- Implemented a Gait-Modeling system with accelerometers data using subspace identification.
Technologies: MATLAB, Java, C, C#