IntelliHire

Inspiration

Hiring the right teachers is critical for educational institutions. This project, IntelliHire, was inspired by the need to streamline the hiring process for educational organizations, ensuring that they can efficiently find candidates whose values, experience, and qualifications align with their organizations' requirements. By leveraging AI to match candidates to job descriptions, IntelliHire provides an efficient and objective tool for assessing fit.

What it does

With a spooky theme for the holidays, IntelliHire analyzes job applications and ranks candidates based on their responses' similarity to a given job description. Candidates are matched with job descriptions using advanced natural language processing (NLP) techniques, resulting in a similarity score that allows employers to see which applicants best meet the criteria. If applicants do meet this criteria, an offer letter will be sent to the applicant's email. This similarity-based ranking saves time, reduces bias, and helps organizations prioritize candidates aligned with their values and requirements.

How we built it

IntelliHire uses React.js for it's frontend and a Node.js and Express.js backend to handle API requests, connect to MongoDB Atlas for candidate data storage, and communicate with Hugging Face’s NLP models for text embeddings. We used the sentence-transformers/all-MiniLM-L6-v2 model via Hugging Face’s HfInference API to generate embeddings for both job descriptions and candidate responses. By calculating cosine similarity between these embeddings, we score and rank candidates based on their fit with the job description. The project integrates Mongoose for database management and is hosted on MongoDB Atlas and uses SMTP for sending emails to applicants.

Challenges we ran into

One of the primary challenges was configuring the AI model and ensuring that embeddings were compatible with similarity calculations. We faced some difficulties with handling and formatting large amounts of candidate data for accurate embeddings and had to manually implement cosine similarity functions to avoid inconsistencies with third-party libraries. Additionally, setting up MongoDB Atlas required careful setup and testing.

Accomplishments that we're proud of

We’re proud of successfully creating an AI-powered solution that leverages NLP to streamline the hiring process. Our accomplishments include integrating real-time AI-based similarity scoring and building a scalable backend that can handle and rank numerous candidates efficiently, updating the mongoDB database accordingly and SMTP services to send offer letter to accepted candidates. It’s rewarding to see how AI can add significant value to a practical application, making a real difference in an educational hiring context as well.

What we learned

Throughout the project, we gained insights into NLP, cosine similarity, and the importance of properly handling and storing embeddings for accurate results. we also learned more about SMTP, and best practices for working with MongoDB Atlas.

What's next for IntelliHire

Future features for IntelliHire includes, to support more sophisticated scoring criteria, such as weighting certain candidate attributes or skills, and analyzing the context and duration of a candidate's experience. Updating the frontend to an interactive dashboard to display ranked candidates with visual analytics and providing interactive tools for recruiters to adjust scoring parameters. This can also be expanded to other job industries as well. These kinds of improvements can make the hiring process more streamlined for recruiters and help find the best people for the job.

Built With

• express.js
• huggingface
• mongodb
• mongodb-atlas
• node.js
• npm
• react