Hi, I'm Gabriel!
Welcome to my portfolio

I'm a full-stack developer with a passion for creating forward-thinking products that have real impact. I'm keenly interest in the world of AI, machine learning, and FinTech, and I'm looking for opportunities to work in exciting teams that make a real difference. Feel free to explore my skills and portfolio below, or scroll up at any point to find my contact links.

Gabriel's Picture
Typescript

Typescript

Python

Python

Javascript

Javascript

SQL

SQL

Bash

Bash

Pytorch

Pytorch

Tensorflow

Tensorflow

Huggingface

Huggingface

STATA

STATA

Supabase

Supabase

Firebase

Firebase

MongoDB

MongoDB

PostgreSQL

PostgreSQL

NodeJS

NodeJS

Prisma

Prisma

Sequelize

Sequelize

Mongoose

Mongoose

ExpressJS

ExpressJS

Koa

Koa

Cypress

Cypress

Jest

Jest

Langchain

Langchain

Redux

Redux

React

React

NextJS

NextJS

Angular

Angular

HTML

HTML

CSS

CSS

Tailwind

Tailwind

Git

Git

Projects

Cadence
Cadence
Repo

Tech Stack

React, Redux, Supabase, LangChain, OpenAI API, Puppeteer

Semantic Querying

Document QA using natural human language for varied file types through use of LangChain, analyzing and organizing disordered data provided using OpenAI's API

Information Scraping

Used Puppeteer to scrape prospect LinkedIn data, allowing for querying of their available information (removed feature)

User Context

Harnessed provided user information to contextualize prompts in order to achieve highly personalized responses and create efficient outreach suggestions. Backend architecture includes Supabase's vector database for document chunking and storage for semantic querying, and document fetching, as well as user and prospect data storage.

GearHub
GearHub
Repo

Tech Stack

React, Redux, TailwindCSS, Express, Supabase, SQL, Stripe, Google Maps API

Real-time Interaction

Implemented persistent real-time interactive rental processes including persistent contracts and chats between users, as well as location-based services for rental search through Google Maps API. Doubled performance of the application by optimizing state management with Redux

Rental Journey

Full user rental journey with authentication using Supabase, calendar setups through a SQL database, and payments with Stripe

Collaborative Process

Managed the workflow of the project on Jira, overseeing and organizing the project's Git utilization and creating outlines of agile sprints for the team

Facial Verifier
Facial Verifier
Repo

Tech Stack

Python, OpenCV, Jupyter Notebook, TensorFlow, HuggingFace

Description

Trained a facial recognition machine learning model using OpenCV, TensorFlow, and HuggingFace to recognize and classify faces. It uses a siamese neural network to achieve this goal, inspired by the architecture described in Koch, Zemel, and Salakhutdinov's paper 'Siamese Neural Networks for One-shot Image Recognition'

Further Development

Next steps are to create a full-stack application to use the model for authentication of users on their own devices, and deploy it into an api that users can train and integrate locally

Transformer Translation
Transformer Translation
Repo

Tech Stack

Python, Pytorch, Huggingface

Description

Created a custom transformer machine learning model based off of the original transformer architecture from the groundbreaking 2017 paper 'Attention is All You Need', which is trainable to be implemented as a translation tool trained on HuggingFace's opus books training set. Currently trained for English to Italian, but can be adjusted dynamically via the config.py file.

Further development

Plans for the future include creating a full-stack educational application to use the model for translation and following its building and training while visualizing its development and training in NLP tasks

Deployment

Please look into the repository for the full code and the model's weights, as well as instructions on how to train this model for different languages locally

TraderML
TraderML
Repo

Tech Stack

Python, Alpaca API, Lumibot, Pytorch

Description

Interactive trading bot that uses machine learning to analyze sentiment and follow a basic strategy to make trades based on its predictions

Trying it out

This bot's trading strategy has been implemented with simplicity in changing and adding to it in mind. Feel free to play around with different strategies and see how your approach compares to the stock of your choice's performance!

Further Development

This project's development is still underway, and it is written to be easily modified and expanded upon. Interactive strategy adjustments have been added, and real-time trading and deployment of this bot is the next step of its development

Domingo
Domingo
Repo

Tech Stack

React Native, Google Places API, Expo, Redux,

Description

Created a mobile application that allows users to more easily find locations catered to their needs and local closure times in Spain through focused use of Google Places API, more easily navigate to these locations, and create lists of their favourite places