The problem Medea solves

Revolutionize Your Resume Screening with AI-powered Chat and Insights

Struggling to sift through mountains of resumes?

We've all been there. This project tackles the challenge of screening resumes more efficiently and effectively for HR professionals.

Here's how it helps:

AI-powered Assistant:
    Chat with a large language model (LLM) to get insights on specific aspects of resumes, like skills or experience.
    Ask clarifying questions and delve deeper into candidate qualifications.

Enhanced Understanding:
    Go beyond keywords.
    Upload resumes and leverage the LLM's ability to process text to uncover a candidate's full potential.

Streamlined Workflow:
    Save time by focusing on the most promising candidates.
    The LLM can handle initial screening, freeing you for in-depth interviews.

Improved Decision Making:
    Gain valuable insights from the LLM's analysis to make better hiring decisions.

This project offers an innovative approach to resume screening, combining the power of AI with a natural chat interface for HR professionals.

Challenges we ran into

Challenge: Combining Text from Chat and PDF Effectively

One of the key functionalities is the ability for the LLM to consider both the user's chat input (focusing on specific aspects of a resume) and the uploaded PDF content (the entire resume). Merging these two sources for a cohesive analysis proved tricky.

Solution: Preprocessing and Concatenation

To address this, we implemented a preprocessing step within the llm_function. Here's how it works:

Extracting Text from PDF: When a PDF is uploaded, libraries like pypdf are used to extract all the text content.
User Query Integration: The user's chat query is then appended to the extracted text from the PDF. This ensures both sources are considered by the LLM.
Contextual Cues: We considered adding techniques like separating the query and extracted text with a delimiter or using special characters to provide context to the LLM. However, in our testing, simply concatenating the text proved sufficient for the LLM to grasp the overall intent and analyze both the user's focus and the full resume content.

This approach streamlined the process and allowed the LLM to effectively analyze the combined information for a more comprehensive understanding of the candidate's qualifications.