Uplink: Internal Company LLM RAG Chatbot platform
I joined the project at an early stage, when it was approximately ten percent complete. I extended the existing codebase to introduce new entities in a non-relational DynamoDB database. I designed and implemented API endpoints following a service–repository architecture and best practices using FastAPI. I collaborated closely with the UX team to improve interface designs and acted as the backend point of reference throughout the project. I implemented vector search functionality and introduced filtering mechanisms to restrict chatbot RAG access, ensuring that users did not share the same level of permissions. I designed and developed a roles and permissions system to control access to content. I worked alongside the DevOps team to improve and modify CI/CD pipelines and deployment workflows. To properly validate backend functionality, I also contributed to frontend development and collaborated directly with frontend engineers. I mentored and guided new developers who joined the project. I managed file upload processes and debugged production issues in AWS, among many other responsibilities associated with the role. I improved query performance by leveraging Python’s asynchronous and concurrent programming capabilities. As the backend lead, I conducted and led technical meetings with clients.
Technologies
Python, Fast API, AWS, Snowflake, Angular, Azure Devops, Github Actions.
Additional Information
This was an internal project developed for Deloitte, accessible exclusively to company employees.
GONAU: All-in-One Platform for Real-Time Delivery Optimization
Worked on a backend system built with NestJS and Supabase relational database. I implemented WebSocket connections to stream real-time user data and order status updates, allowing managers to monitor each order’s state and assignment in real time. I developed REST endpoints to retrieve and delete users, and integrated Auth0 APIs to access user data and enforce role-based access control across different endpoints. Additionally, I created duplicate user records in Auth0 and implemented logic to merge and align user data from Auth0 with data stored in Supabase to ensure consistent access management.
Technologies
Nest JS, Node JS, React, Flutter.
Additional Information
Private repo for business reasons.
SwaplyAr Transaction Platform
Worked collaboratively with the data engineering team to design and develop backend endpoints using Express. I implemented endpoints to retrieve user information, delete users, add payment methods, and contributed to the authentication logic based on JSON Web Tokens (JWT). I also collaborated on the design of database tables related to users and their associated accounts, among other components.
Technologies
Node JS, Express, SQL.
Additional Information
The system was fully functional and the authentication flow was operating correctly during development and deployment. At a later stage, issues were reported that appear to be related to changes or misuse on the client side, which impacted the authentication behavior. At the time of delivery, the authentication system was stable and working as expected.
Mechanical Workshop Website
Full-stack Django template for automotive services. The app allos to add users sing the Django Admin. Django template tags iterate over each car object on the database and displays on the page.
Technologies
Python, Django, PostgreSQL, Docker, Bootstrap, HTML, CSS, JS, Jinja2, Linux.
Patients API
REST API for patient data management with PostgreSQL. The API has JWT authentication on the endpoints. It can create, delete, update and get patients data following rest architecture.
Technologies
FastAPI, PostgreSQL, Python, Docker, JWT.
Additional Information
Personal project developed to explore and gain hands-on experience with FastAPI.
Real State Contracts App
Leasing properties app orchestrator built with Flutter. The app allos to login an send real state contract invitation to other users and check the status of each step of the contract.
Technologies
Flutter, Google Cloud, Firebase, NoSQL database.
Additional Information
Private repo for business reasons.
BLK Market Chatbot
Web-based customer service chatbot powered by ChatGPT. I had to add functions to get real time data from the web page to answer cleint's questions and do prompt engineering to make the chatbot more accurate.
Technologies
Next JS, TypeScript, MongoDB, Chat GPT API.
Additional Information
The solution was fully deployed and operational in production, running on a Heroku server using Docker containers. It integrated real-time data retrieval from the web application, authentication, data processing, and external API integrations. The feature was later discontinued due to budget constraints.
Sync Data AI Compliance Platform
AI Platform for company compliance document processing. The platdform is used to extract text from PDF files and create reports to check if they are aligned with certain norms such as ISO regulations. The project started as a list iteration and ended up being a step functions flows that would creat AWS Lambda functios by demand and create objects in Wordpress API to create reports.
Technologies
Python, AWS (Lambda, S3, KMS, Step Functions, API Gateway, Elastic Beanstalk), Open AI API, WordPress, WordPress API, PHP, Flask
Additional Information
Private repo for business reasons.
Home Banking Desktop App University Final Project for Programming 3 assignment
Java Swing application with PostgreSQL database integration. The app allows to login as administrator or client and execute actions such as send money to other user, create user, see user data, etc. This was the final project for the Programming 3 assignment at the University of Buenos Aires. Where I learnt a lot about Java, SOLID principles, Design Patterns, and Object-Oriented Programming.
Technologies
Java, Swing, Docker, PostgreSQL, Linux.
WhatsApp Chatbots powered by Open AI API
AI-powered chatbots for customer service, payments, and order tracking. Tool calling implementations allows Open AI api to interact with external databases and apis to provide real time data to the client.
Technologies
Python, Flask, FastAPI, TiendaNube API, MercadoPago API, WhatsApp API, GraphQL, MongoDB, Large Language Models (Anthropic, Open AI API), Linux, Strapi.
Additional Information
Project developed for an early-stage startup. The product was discontinued due to budget constraints; however, the chatbots were fully functional and deployed on a Heroku server, running inside Docker containers, with authentication, data processing, and integrations with external APIs.