Kevin – Python, Django, FastAPI

Ready-made integrations for marketing workflows

• Designed the software architecture;
• Implemented the whole back end based on Actix-web on the back end and NextJS on the front end;
• Hired front-end devs to continue implementing the front end;
• Made the app handle 100s concurrent requests per second with a single instance;
• Implemented the whole project in AWS.

AI chatbot to handle Shopify support requests for B2C stores.

• Rebuild from scratch the whole integration with OpenAI's models;
• Developed a layered structure to handle different kinds of conversation classification;
• Integrated it with Gmail using Google APIs.

This project was developed for a Peruvian institute, which consisted of integrating their Chat engine with OpenAI's ChatGPT, to handle conversations with potential students.

• Planned the whole architecture of the project and developed from scratch;
• Created an integration with ChatGPT's API and Chatra, handling conversation histories and caching;
• Created an integration with Google Docs to edit the prompt in real-time.

US startup backed by YC, with a platform that lets you find and easily apply for jobs.

• Was in charge of the scrapers, handling data from thousands of sources;
• Handled the migration to SQLAlchemy v2;
• Implemented a NLP parser using OpenAI's GPT-3.5 model;
• Optimized our queries to PostgreSQL to reduce the load and I/O time.

The product is developed for a SaaS company focused on attendance management and communication between manufacturers and hourly workers. Worked on several projects, including a migration from Postgres to ElasticSearch, and a dashboard called Attendance Insights that shows statistics about the company's historical attendance data.

• Joined the team to focus on the Django app, but also had the chance to work with React and Typescript;
• Orchestrated a workaround for our CI/CD pipeline, integrating GitHub Actions and CircleCI;
• Lead a project to provide the customers with historical analytics, planning the database architecture, ETL process, and building the API in a way that it was fast and scalable;
• Implemented OpenAI's GPT-3.5 to an incoming messages listener, with a layers architecture to save costs;
• Migrated a huge Postgres table to a micro-service that writes to ElasticSearch;
• Was part of the team in charge of the audits/tests in order to obtain the SOC2 certification.

The company started as a FastAPI app to handle complex integrations (those that were impossible to create with Zapier/Workato), for several clients. After a while, the project scaled and the we decided to migrate from FastAPI to Golang and Gin, in order to handle concurrency in a more efficient way. Currently the project handles millions of requests per day.

• Developed the first version using FastAPI;
• Abstracted the integrations and API objects to reuse them across different clients;
• Integrated dozens of APIs using HTTPX async library to make non-blocking API calls and maintain the app's ability to handle concurrent requests;
• Moved away from FastAPI and re-wrote the app with Go and Gin;
• Moved away from Heroku to Virtuozzo.

A housing startup that rents luxury rooms and apartments, giving guests the experience of a luxury hotel with the comfort of being at home. The website and app allow guests to book and manage their bookings, and the House staff provide the guests with services and support from the moment of validating their booking to manage the guests' requests during their stay.

• Developed an Apple Wallet integration to enable our hosts to have a smoother check in experience.
• Worked on optimizing the website speed by caching the content and refactoring Django’s DB queries.
• Developed a GraphQL API from scratch for the mobile app, using the library Graphene.
• Bug fixes throughout the whole Django App.
• Created a verification process using the user's camera.

Software as a service that helps social media managers generate leads from Facebook comments and messages by quickly answering their questions, classifying them, and collecting their contact info. It evolved with time to allow more functionalities (rules, webhooks, NLP) but ended after not generating enough money to be a profitable business.

• Developed and integration with the Facebook API
• Developed the API using Graphene/Django
• Created microservices to execute asynchronous tasks, triggered via Cloud Tasks
• Did some beta MVPs integrating our app with Wit.ai to implement NLP.