Inspiration
The inspiration for this project comes from the challenges autistic individuals often face with the repetitive and mentally taxing process of filling out online forms. This extension was designed as a practical tool to automate this task, specifically to streamline job applications and lower barriers for anyone who finds the process overwhelming.
What it does
AI Form Filler is a Chrome extension that automates form completion. A user pastes their resume into the extension's interface and clicks a button. The extension then analyzes the input fields on the current webpage and uses an on-device AI model to intelligently map the resume data to the correct fields, populating the form automatically for the user to review and submit.
How we built it
The extension is built with standard web technologies (HTML, CSS, JavaScript) and integrates an on-device Large Language Model. It works by identifying all form inputs on a page, packaging them with the user's resume, and sending this data to the AI. The model then returns a structured JSON object mapping the correct information to each form field. Due to hardware limitations preventing the use of Gemini Nano, this proof-of-concept uses meta-llama/Llama-3.2-3B-Instruct to demonstrate the full functionality.
Challenges we ran into
The primary technical challenge was the lack of a GPU capable of running Gemini Nano on my development machine. To overcome this, I substituted a comparable on-device model to ensure the core concept could be built and tested effectively. This required focusing all development effort on the core AI and data-mapping logic rather than on UI polish.
Accomplishments that we're proud of
I am proud of successfully creating a fully functional, end-to-end proof-of-concept that solves a real-world problem. The key accomplishment was successfully integrating an on-device LLM to intelligently parse unstructured resume text and map it to structured web form fields, all within the constraints of a browser extension.
What we learned
This project provided deep, practical experience in the architecture of AI-powered Chrome extensions, particularly concerning the use of on-device models. I gained valuable insights into prompt engineering techniques required to reliably extract structured JSON data from free-form text, which is the technological core of this tool.
What's next for AI Form Filler with Gemini Nano
The roadmap is focused on growth and refinement. The immediate next step is to integrate Gemini Nano as the core on-device model. Following that, priorities include a complete UI/UX overhaul for a more polished user experience, adding features like multi-page form support and saved user profiles, and expanding its purpose as a broader accessibility tool for all users.
Log in or sign up for Devpost to join the conversation.