Conducted research in Life Cycle Assessment, predicting impact and life cycle inventory through chemical structure.
•
Developed the CLiCC tool (https://clicc.net/), which helps LCA researchers to estimate chemical life cycle impact through web UI.
•
Developed groups of modules, including ANNs using Tensorflow, in Python to estimate the environmental impacts of chemicals from different perspectives;
•
Created RESTful APIs in Flask and Django to wire-up the Python modules with a front-end in React;
•
Developed the backend which handles multiple user requests and long tasks asynchronously using uWSGI, Celery and RabbitMQ. Wrapped-up services in Docker and deployed it on AWS with docker swarm;
Developed a web platform, Michaniki, that allows user to use, build, and retrain image classifiers using custom images for any classes;
•
Conducted transfer learning and fine-tuning on a pre-trained Inception V3 CNN model in Tensorflow, Keras and Celery. Built a stable inference sever using Redis and Flask that can handle massive user requests;
•
Wrapped-up all services of Michaniki using Docker. Made it easy to deploy.
