Background pattern

Hi, I'm Matt

AI/ML Focused Software Engineer

Cornell Tech CS Master’s student and Full-Stack Engineer. I specialize in deploying high-impact ML solutions to solve real-world data challenges.

Matthew Shapiro
Beach Background

Technologies I Use

Python
Python
SQL
SQL
TypeScript
TypeScript
JavaScript
JavaScript
Java
Java
Ruby
Ruby
C
C
HTML
HTML
CSS
CSS
Swift
Swift
Python
Python
SQL
SQL
TypeScript
TypeScript
JavaScript
JavaScript
Java
Java
Ruby
Ruby
C
C
HTML
HTML
CSS
CSS
Swift
Swift
sample
Large Language Models (LLMs)
sample
Retrieval-Augmented Generation (RAG)
sample
PyTorch
sample
Transformers
sample
Fine-tuning
sample
Vector Databases
sample
LangChain
sample
Natural Language Processing (NLP)
sample
Transfer Learning
sample
Next.js
sample
React
sample
Node.js
sample
Ruby on Rails
sample
Tailwind CSS
sample
PostgeSQL / NoSQL
sample
RESTful APIs
sample
Cloud Migration
sample
AWS (Certified Cloud Practitioner)
sample
Github Actions
sample
Infrastructure as Code (IaC)
sample
Docker
sample
CI/CD Pipelines
sample
Automated Testing (PyTest/Jest)
sample
ETL Pipelines
sample
AWS Lambda
sample
Amazon S3 & EC2

My Experience

Cornell Logo

Cornell Tech

Master of Engineering

date2025 - 2026

Areas of Study:

Deep Learning, Applied Machine Learning, Data Science, HCI

Colgate Logo

Colgate University

Bachelor of Arts

date2021 - 2025

Relevant Coursework:

Software Engineering, NLP, Database Systems, DSA


Johnson & Johnson

Johnson & Johnson

Platform Operations Engineer Intern

locationNew JerseydateSummer 2025

Designed, built, and deployed a GenAI support agent using LLMs, RAG architecture, and vector databases.

Principal Financial Group

Principal Financial Group

Full Stack Software Engineer Intern

locationCharlotte, North CarolinadateSummer 2024

Developed and migrated full stack applications to the cloud utilizing AWS, Next.js, TypeScript, and CI/CD DevOps deployment techniques.

Bridgify

Bridgify

Back End Developer Intern

locationTel AvivdateSummer 2023

Utilized Python and AWS to efficiently retrieve and process data from multiple sources, optimizing data handling and storage throughout the ETL pipeline.

background

Projects

RetinaLiteNet: Medical Image Segmentation & Disease Classification

Developed a hybrid CNN-Transformer architecture in PyTorch for retinal analysis. Achieved a 0.755 Dice score and 98.3% specificity on segmentation while reaching 90% accuracy for 45-class disease classification.

PyTorchTransformersComputer Vision
RetinaLiteNet: Medical Image Segmentation & Disease Classification

Next-Query Prediction: Comparative Sequence Modeling

Systematic study of paradigms including Markov chains, ALBERT+GRU, and T5 models. Fine-tuning a T5 transformer outperformed deep learning models trained from scratch by over 15x.

Fine TuningPredictive ModelingNLP
Next-Query Prediction: Comparative Sequence Modeling

Automated Toxic Comment Classification Model

Designed a feed-forward neural network using Python and TensorFlow to identify harmful content. Optimized architecture and training to achieve a 92% accuracy rate in detecting identity-based hate.

TensorFlowSupervised LearningNeural Networks
Automated Toxic Comment Classification Model

GameShelf: Video Game Ranking SaaS Application

Engineered a full-stack SaaS application using Ruby on Rails and JavaScript. Implemented a responsive front end with TailwindCSS and integrated secure user authentication with a relational database.

Ruby on RailsRelational DatabasesFull Stack
GameShelf: Video Game Ranking SaaS Application
Contact Background

Contact Me