- Collaborated with team members to research and create an LSTM model that predicts future loads for ECS clusters given past data in order to generate recommendations for the number of stacks to be allocated in the future for a given cluster
- Optimized cloud costs using the recommendations from the LSTM model to reduce expenses spent on idle clusters
- Programmed a REST API using Flask and Python to allow users to easily run the machine learning model with a URL
- Analyzed trends in data collected on the loads for ECS clusters using pandas within Jupyter Notebook
- Contributed to writing a Dockerfile to run model in an AWS cluster and connect to an RDS database and Snowflake
- Extracted information from databases in Snowflake and stored database credentials in a Hashicorp Vault using REST APIs
- Interacted with AWS resources such as lambdas, RDS databases, and ECS clusters