Inspiration
The job market today is tough. Candidates are applying everywhere, often settling for roles that aren’t the best fit for their skills or aspirations. On the other hand, companies are overwhelmed with thousands of applications and often end up rejecting candidates without giving them a fair chance. The result frustration on both sides.
What it does
SmartApply solves this by introducing an intelligent pre-screening process that identifies the most suitable candidates for a role before the interview stage. The platform uses a combination of:
- Interactive voice-based interviews where an AI bot engages with the candidate, evaluates their responses, tone, and communication style.
- Smart scoring that assesses candidates on three dimensions, job description match, resume quality, and conversational performance, ensuring that recruiters see the right candidates first.
- AI-powered resume analysis to match candidates to job descriptions beyond keyword-based filters.
How we built it
We built an AI agent that intelligently evaluates different sections of a job description and compares them with key elements extracted from a candidate’s resume. To go beyond traditional ATS systems, we incorporated contextual embeddings and semantic similarity models instead of plain keyword matching. Additionally, we integrated a real-time conversational AI module that interacts with candidates, analyzes their spoken responses, and assesses soft skills like confidence and communication.
Challenges we ran into
One of the biggest challenges was capturing human empathy and behavior through AI interactions. Understanding tone, hesitation, or enthusiasm in speech required fine-tuning emotion detection and sentiment analysis models. Another challenge was ensuring fairness and bias reduction in scoring candidates with diverse communication styles and backgrounds.
Accomplishments that we're proud of
Successfully fine-tuned an open-source model for multi-purpose evaluation:
- Sentiment analysis
- Candidate assessment
- Profile summarization Created an end-to-end automated screening pipeline that can handle both resume parsing and live AI interviews. Developed a reliable scoring system that improves hiring efficiency while maintaining transparency for both candidates and recruiters.
What we learned
We learned how challenging yet impactful it is to bridge human intuition with machine intelligence. Working on this project deepened our understanding of AI model fine-tuning, speech emotion recognition, and real-time interaction systems. It also reinforced the importance of designing AI systems that are fair, interpretable, and human-centric.
What's next for SmartApply
Next, we plan to:
Build a dashboard for recruiters to visualize candidate rankings and insights.
- Integrate voice cloning and dynamic questioning, allowing the AI interviewer to adapt in real time.
- Introduce feedback loops to continuously improve model accuracy based on recruiter evaluations.
- Launch SmartApply for universities to help students prepare for interviews and get personalized career insights.
Log in or sign up for Devpost to join the conversation.