Promaxo, Inc. is a series-B medical device start-up specializing in early cancer detection

Responsibilities include machine learning system development, infrastructure/devops, and product development for surgical robot arms:

• Developing U-Net-based deep learning models in PyTorch for medical image segmentation (rectum/lesion) and classification

• Reducing inference latency from 60s to 40s for model serving pipeline on GCP

• Writing and maintaining Bash scripts to perform quarterly upgrades for 40+ production systems

• Migrating critical image processing algorithms from C++ to Python

• Developing testing and controlling interfaces for surgical robot arms

CS 325: Algorithms and Data Structures (E-campus)

• Implement proof-of-concept experiment for 3 published papers in applied machine learning and privacy in machine learning

• Develop data poisoning methods that increase the risk of privacy leakage from deep learning models by 8 times with researchers from Google, Northeastern University, and National University of Singapore.

• Create a machine learning solution to apply Hidden Markov Model to extract the playing and walking patterns from 12-month-old infants to apply to robotic exploration with researchers from NYU and University of Utah.

• Design and deployed Microsoft PowerBI reports for pharmaceutical sales teams.

• Engineered deduping and recording matching tools for internal usage.

• Attained data from data warehouse, cleaned, and prepared data to perform segmentation and

territory alignment analysis for pharmaceutical sales.

• Performed QC on deliverables before shipping to customers.

• Implementing different supervised and un-supervised machine learning algorithms from cutting-edge research to target new customer bases for various pharmaceutical products.