Project Proposal

Matthew Martin

Vision Statement:

This project will allow me to gain knowledge in new tech stacks, with a focus on the devops aspects of software engineering and machine learning. It will help me build familiarity with new tech tools like java, spring boot, docker, and AWS, while simultaneously connecting my biology background with my computer science education. —

Motivation:

This project matters because I have been focusing on a niche area of computer science in order to be a more competitive applicant, this area being Bioinformatics/Healthcare Software Engineering/Biotech AI. This project involves expanding my knowledge in these specialities so that I can better secure these roles but also perform well in them. I will learn all the technologies I want to learn , gain experience applying ML in a domain - relevant context, and create something that is industry relevant. —

Goals Timeline and Deliverables:

Core API Development (weeks 1-3)

Create a Spring Boot application with biological data models.
Implement CRUD operations for patient data, biological samples, etc.
Add user authentication and authorization.
Set up PostgreSQL database.
Implement error handling.

Containerization and deployment(weeks 4-5)

Dockerize the application
Create docker-compose for local development
Deploy to AWS

ML integration(weeks 6-7)

Implement a simple NLP model that will classify data for HIPAA compliance
Use HuggingFace for the model
Create a REST endpoint in the SPring Bior app to call the ML service

Advanced Features and Portfolio (weeks 8-10)

Create detailed documentation
Create a simple frontend to showcase the API
Testing and validation

Risks:

The risks involved with the project stem from time management and unfamiliarity. Learning new languages and tools involve a significant learning curve, and with about 11 weeks to have a deliverable project, the amount of time may be compressed, with only 45 hours available for the project. This is an ambitious project with many different aspects with devops, machine learning and RESTful API’s being distinct parts of software engineering, so there is little overlap between them, meaning aspects of one does not provide much background on the other.

Mitigation:

To successfully deliver this project, I am adopting a learn as I go style. This will prevent me from getting hung up on certain concepts or delving too deeply into complex aspects of a language or technology that may not be relevant to completing the project. This integrated learning strategy means that I first learn enough Java to create my classes , and focus specifically on the topics necessary for building an API. Then when I am integrating AWS, focusing on the services like EC2,RDS,and S3 that are directly applicable to my project and not delving into other AWS services deeper than necessary. This will keep me on schedule and avoid undue delays.

Criteria:

My criteria for project completion are the same as outlined above, if my project contains all the deliverables listed then it will be considered successful. That being said to be 100% it should pass all testing and evaluation. It should be able to run docker compose (configure the project docker successfully) , send a clinical note to /upload(sucessful API), get a de-identified version stored in the DB (ML model recognizes and protects protected healthcare information), and retrieve a compliance report that shows was PHI was detected.

Project Website:

The semester-long project is part of my general portfolio. There is a direct link to the project portfolio, and it can also be accessed through my general portfolio github page as well.

https://matthewmartin117.github.io/project/index.html