I applied for an internship at Mistral AI and one of the requirements for the application was to build a project using either their API & SDK with Next.js, or anything in Python using vLLM. I opted for the Next project, and built a simple web UI for the Mistral AI API.
It’s really just a simple Next.js app styled with Tailwind CSS and shadcn/ui that allows you to interact with Mistral models (currently mistral-small-latest). It supports Markdown and LaTeX rendering, code blocks with syntax highlighting, and saving chats locally in your browser’s IndexedDB thanks to Dexie.js.
Building this project was a really fun experience, since I had never worked with LLMs before, nor interacted with IndexedDB. I’m really glad Mistral’s job offer required me to build this, because I learned a lot from it and quickly got familiar with their API. I feel like more companies should do this (no need for it to be a big project, just simple stuff to make candidates interact with the company’s products).
Anyways…
Basically the fundamentals of LLM APIs and data streaming, and local data storage with IndexedDB. Overall a fun little project I’m really proud of despite its simplicity, 10/10, would recommend. Go try it out here! You will simply need a Mistral API key, which you can get for free by signing up on their website.