Ben Hebert

Mentor AI allows users to upload their journal, diary, notes, or thoughts either as raw text or (coming soon) as word, pdf, or txt files. A background worker will then review their uploaded notes using ChatGPT to provide them with personalized guidance in the form of a weekly email according to the preferences that the user sets. This app uses Vite on the frontend and a Firebase backend with Celery and Redis for handling the weekly email tasks.

I helped Winncom Technologies with their first forray into automated order processing by leading a project to install and configure an ecommerce platform that would integrate with their existing ERP system. I utilized SQL to interact with the ERP database and HTML+CSS to design the frontend of the site according to the client's preferences.

FolderLab allows users to upload files or folders containing Word or PDF documents and ask those files/folders questions (I built this before the major LLM companies offered this service!). Building this app helped me practice extracting raw text from documents, using Langchain retrieval methods, and working with Pinecone to store the embeddings of the documents.

In a quick 2-week engagement, I helped drastically improve the appearence and search functionality of the Wired Nomad's website. I used vanilla HTML, CSS, and Javascript to redesign the site and worked with Firebase to query the listings data. See the before and after here.

Very similar to FolderLab, DataRoomDog allows users to upload files or folders containing Word or PDF documents and ask those files/folders questions. I built this app to practice using the Chroma vector database, Langchain, Next.js and Supabase.

Automates the legal and administrative work of a 'Qualified Intermediary' in a 1031 exchange. Allows users to input information about their real estate properties which gets automatically mapped into the relevant legal contracts for 1031 exchanges. Admin capabilities include assigning tasks to employees, adding additional information into contracts, and releasing contracts for e-signaturt. Built with Next.js, Django Rest Framework, Sendgrid, and service called RightSignature for e-signatures.

One of my first projects to ever build, this Django app mixes a cellular data strength map with common Google Maps searches so digital nomads pick their next location with confidence. I ran Django scripts in the Heroku cloud to scrape nearly 10 million cell tower coordinates from OpenCellID into a Postgres database. This project helped me practice writing clean, modular Javascript code and bundling it for production with Webpack.