Project Title

Smart Hire AI Agent: Intelligent Resume & Interview Assistant

Abstract

As college students, we face two critical challenges while applying for jobs: ensuring resumes pass through Applicant Tracking Systems (ATS) and preparing effectively for company-specific interviews. Many talented candidates are rejected early due to poor alignment with job descriptions, missing keywords, or insufficient interview preparation.

To address this, we propose Smart Hire AI Agent, an AI-powered solution leveraging IBM’s Granite models and Agent Development Kit (ADK). Unlike existing platforms that either optimize resumes or provide generic interview practice, our system integrates end-to-end employability support:

1. ATS-Friendly Resume Analyzer – evaluates resumes against specific job requirements, assigns a competitiveness score, and highlights missing skills.
2. Personalized Mock Interview Generator – creates role- and company-specific interview simulations, including HR and technical rounds, with both text and voice interaction.
3. Recommendation Engine – suggests targeted courses, certifications, and projects to close identified skill gaps.

What makes our approach novel is the fusion of agentic workflows and Retrieval-Augmented Generation (RAG), enabling company-specific, adaptive feedback that scales across multiple roles. By combining text and voice-based simulations, the system creates a realistic interview experience while remaining modular and extensible for diverse hiring contexts.

The expected outcome is an AI employability assistant that not only improves candidate readiness but also bridges the gap between academic preparation and industry hiring needs, empowering students to succeed in competitive recruitment processes at scale.

(a) Solution Approach

The system follows a multi-layered approach to prepare students more effectively for hiring. (1) It uses NLP-based resume parsing and ATS benchmarking to extract skills, experiences, and keywords, while evaluating resumes against role-specific job postings. (2) A Retrieval-Augmented Generation (RAG) pipeline builds a knowledge base from resumes, job descriptions, and interview question banks, enabling the AI to generate tailored company- and role-specific mock interview sets. (3) An agentic workflow powered by IBM ADK automates resume scoring, feedback generation, and mock interview scheduling, reducing manual effort and ensuring consistency. Finally, the system integrates voice-enabled interview simulations, where IBM Watson Speech-to-Text transcribes candidate responses, and Granite models score the answers for clarity, confidence, and relevance.

(b) Tools, Libraries & Datasets

The project integrates IBM Granite LLMs for semantic similarity and NLP tasks, IBM ADK for orchestrating intelligent agents, and IBM Watson Speech-to-Text for voice-based interview experiences. To train and evaluate the system, datasets include public job descriptions (sourced from LinkedIn and Naukri), open-source interview question banks, and sample resumes (2–3 per company/role), ensuring realistic role-specific benchmarking and preparation.

(c) Expected Outcome

The expected result is a smart AI-powered assistant that directly addresses three pain points:

1. Improving resume alignment with ATS filters by highlighting missing skills and keywords.
2. Providing adaptive mock interviews tailored for both HR and technical roles.
3. Recommending personalized learning resources such as online courses, certifications, or projects to bridge skill gaps. Together, these features reduce preparation time while boosting the hiring success rate of students.

Built With

• adk
• fastapi
• flask
• ibm-cloud
• ibm-granite
• langchain
• python
• pytorch
• rag-pipelines