AI Resume Coach

• 

  UI for AI Resume Coach

Inspiration

Every student in the world wants one thing: a job. But like many of my peers, I feel uncertain about whether the skills I have are translated and explained well in my resume. This inspired me to create an AI-powered Resume Coach to provide instant, personalized feedback that would help job seekers improve their resumes and feel more confident when applying for jobs.

What it does

The AI Resume Coach allows users to upload their resumes in PDF format and receive instant, AI-powered feedback. The app analyzes the content of the resume and compares it to others in similar fields. Using advanced natural language processing and machine learning models, it provides actionable suggestions focused on improving formatting, wording, and content to better align the resume with job goals. The goal is to help job seekers optimize their resumes, boost their chances of landing interviews, and feel confident in their applications.

How I built it

The app was built using Streamlit for the interface, which allowed me to quickly create a simple and intuitive web application. I then took a dataset of 13,000+ resumes and preprocessed them using Natural Language Processing. I added this preprocessed data into a database using ChromaDB. These stored resumes act as a "training set" for the AI model to look at stored resumes for context. For the AI-powered feedback, I used Hugging Face’s GPT-2 model to generate detailed feedback on resumes. From the user's POV, the process starts with a user uploading their resume, then the app extracts text from the PDF, processes it, and uses the AI model to generate feedback, which is displayed to the user in an easy-to-understand format.

Challenges I ran into

One of the major challenges I faced was integrating PDF extraction into the app. Extracting text from PDF resumes in a way that preserved the formatting and structure was tricky, especially when dealing with resumes of varying complexity. I also had to refine the AI feedback generation to ensure that the suggestions were practical and actionable without being too generic. Performance was another concern, as generating feedback could take a few seconds, and I wanted to balance speed and accuracy.

Accomplishments that I'm proud of

I'm proud of building my first functional app that allows job seekers to receive personalized resume feedback in real-time. The integration of GPT-2 for text generation provided high-quality, context-aware feedback, and ChromaDB helped me store and retrieve similar resumes for comparison. Deploying the app on Streamlit Cloud made it easy to share with users, making it accessible to anyone who needs it. I'm especially proud of how the app supports students and job seekers who might not otherwise have access to professional resume feedback.

What I learned

This project taught me a lot about integrating machine learning models with web applications. I learned how to use Hugging Face’s model pipeline and Streamlit to quickly prototype a functional app. I also gained hands-on experience with ChromaDB for managing and querying embeddings, which was essential for making the feedback more relevant. Most importantly, the project reinforced the value of user experience in building an app that not only works well but is also intuitive and easy to use.

What's next for AI Resume Coach

Maybe after this competition, it's time to get a team together and create a more fleshed-out application. This project has a lot of potential to help any college student.

Built With

• chromadb
• hugging-face
• natural-language-processing
• python
• streamlit
• vscode