Abstract

This paper presents an AI-powered chatbot designed to assist HR managers in screening developer resumes efficiently. The system leverages LangChain, FAISS, and OpenAI technologies to process, analyze, and retrieve relevant candidate information from resumes. By automating the initial screening process, the chatbot reduces manual effort, minimizes biases, and enhances decision-making for recruiters.
Additionally, the chatbot enables recruiters to assess candidates holistically, evaluating not only technical skills but also career consistency, leadership potential, and adaptability. It can also ask probing questions that are difficult to address in traditional interviews, ensuring a more comprehensive candidate evaluation.
Results indicate a 6x reduction in resume screening time, 30% fewer unnecessary interviews, and an estimated 38% reduction in hiring costs.

Introduction

Recruitment is a time-intensive process, with HR professionals spending 50+ hours per job opening on screening resumes and conducting interviews. Traditional methods are prone to biases, inefficiencies, and language barriers. AI-driven solutions provide a scalable, unbiased, and data-driven approach to hiring. Our chatbot automates resume screening, allowing recruiters to prioritize high-potential candidates and reduce early-stage hiring inefficiencies.

Problem Statement

The traditional hiring process often involves manual resume reviews followed by initial interviews, where language barriers and subjective biases may affect candidate evaluation. Recruiters spend significant time conducting first-round interviews just to determine if an applicant is worth further consideration.

Objective

This project introduces an AI chatbot that enables recruiters to interact with resumes and cover letters just like they would with a candidate. Instead of conducting an initial interview, recruiters can ask questions about an applicant’s skills, experience, and suitability, allowing AI to analyze the resume content and provide insights instantly.

The chatbot’s primary goal is to:

• Minimize unnecessary first-round interviews by filtering candidates early.
• Reduce human bias by focusing on skills and experience rather than language fluency.
• Increase efficiency by allowing HR managers to interact with resumes via AI instead of scheduling multiple
  calls.
• Enable deeper candidate evaluation by allowing recruiters to ask probing questions about career consistency, mental strength, and adaptability factors that are difficult to assess in traditional screening.

Background

Companies today face challenges in hiring the right talent efficiently. According to a LinkedIn report, 75% of resumes are never seen by a human due to ATS (Applicant Tracking Systems) filters. Additionally, 30-50% of hires fail within the probation period, leading to higher recruitment costs and productivity losses.

Current Gaps in Hiring Practices

Despite advancements in AI-driven recruitment, many hiring processes still rely on:

• Keyword-based filtering rather than context-aware candidate evaluation, causing skilled applicants to be overlooked.
• Time-consuming manual screening, which slows down the hiring cycle.
• Limited ability to assess soft skills such as leadership, adaptability, and problem-solving.
• High failure rates in probation due to misaligned hires and lack of predictive insights.

Industry Trends & Market Data

• AI-driven hiring tools reduce hiring time by 30% and improve candidate matching.
• Misaligned hires cost businesses an average of 20,000 in wasted salaries and recruitment expenses.
• Companies using AI for resume screening have a 60-70% higher probation success rate compared to traditional methods.

This project addresses these gaps by leveraging AI-powered resume screening, NLP, and retrieval-augmented generation (RAG) to enhance hiring efficiency and reduce recruitment risks.

Methodology

1. PDF Document Processing

• Extracted and cleaned text from resumes and cover letters using PyPDFLoader.
• Preprocessed text for analysis, ensuring better candidate insights.

2. Text Embedding & Vector Storage

• Generated embeddings using HuggingFace's BAAI/bge-small-en-v1.5 model.
• Stored embeddings in FAISS for quick retrieval.
• Add metadata (e.g., job role, skills, experience).

3. Text Chunking for Enhanced Query Accuracy

• Implemented RecursiveCharacterTextSplitter to divide large text into smaller, context-preserving chunks for efficient processing.
• Improved query accuracy by enabling precise retrieval of candidate information during resume screening.

4. AI-Driven Chatbot for Resume-Based Q&A

• Developed a FAISS-based retriever for semantic search within resumes, enabling quick and accurate information retrieval.
• Integrated ChatOpenAI to generate dynamic responses based on resume content.
• Designed custom prompt templates to allow recruiters to ask targeted questions, such as:

1. Who can do the best job in Python development?
2. Who has experience working with fintech companies?
3. Which candidates have demonstrated leadership or project management skills?

5. AI-Driven Candidate Pre-Screening

• Enabled recruiters to query resumes before conducting interviews.
• The chatbot evaluates not only technical skills but also soft skills like leadership, mental resilience, and career consistency, enabling recruiters to identify candidates with long-term potential.
• Helped businesses filter out unqualified candidates without language barriers or subjective bias.

6. Environment Setup and API Integration

• Configured OpenAI API for AI-driven responses.
• Installed and managed essential Python libraries, including LangChain, FAISS, and HuggingFace Transformers.

Results & Impact

Our chatbot significantly improves recruitment efficiency and reduces hiring costs. Key findings:

Conclusion

The AI-powered chatbot streamlines resume screening, reduces hiring inefficiencies, and enhances probation success rates. By leveraging NLP, vector databases, and OpenAI, the system enables recruiters to make data-driven hiring decisions while minimizing biases and operational costs.
Additionally, by focusing on holistic candidate assessment and long-term fit, the chatbot not only optimizes hiring but also reduces the risk of costly mismatched hires, ensuring higher probation success rates.

Key Technologies and Tools

• LangChain for LLM integration.
• FAISS for vector search.
• OpenAI API for language model responses.
• HuggingFace Models for text embeddings.
• PyPDFLoader for document processing.
• Python for end-to-end development.

Future Work

• Implementing real-time candidate scoring.
• Enhancing chatbot multi-language capabilities.
• Integrating with existing HR software for seamless workflow