ScreenSmart
Fast, Fair, and Flawless Hiring – Powered by AI.
Created on 5th April 2025
•
ScreenSmart
Fast, Fair, and Flawless Hiring – Powered by AI.
The problem ScreenSmart solves
ScreenSmart reimagines the resume screening process by eliminating bias and inefficiencies that plague
traditional recruitment. Leveraging AI, NLP, and Machine Learning, it introduces precision and fairness into
hiring.
Resumes are processed using a SpaCy-based NLP pipeline combined with an LLM-powered evaluator for
deep contextual understanding.
The platform generates:
- ATS Compatibility Score to improve applicant visibility
- SWOT Analysis highlighting candidate strengths and weaknesses
- Match Percentage using:
- TF-IDF for keyword extraction
- Cosine Similarity for measuring relevance
- Skill Graphs for context-aware skill mapping
To forecast hiring outcomes, we built a Predictive Hiring Model using Extreme Gradient Boosting, trained on
500 candidate profiles. The pipeline includes:
- SentenceTransformers (MiniLMv6) for semantic embeddings
- TF-IDF, Feature Scaling, and One-Hot Encoding for preprocessing
- This model achieved 76% prediction accuracy for hiring success.
Additional features:
- Candidate Comparison Reports with visual metrics
- Built-in Email Notifications for seamless communication
Overall, ScreenSmart is a comprehensive tool for smarter, faster, and bias-free hiring decisions
Challenges I ran into
Challenges I ran into
Building ScreenSmart during a 24-hour offline hackathon was intense, to say the least.
Initially, we opted for Supabase to handle storage and authentication. Things seemed smooth—until we hit unexpected issues with querying and role-based access. After wrestling with the bugs for hours, we finally got it running.
But then came the curveball: a mentor pointed out that the candidate-facing functionality didn’t align with the hackathon prompt. So, several hours of work had to be scrapped on the spot. That was a hard lesson in fully aligning with the problem statement before diving into development.
Still, we adapted quickly, shifted focus, and managed to deliver a clean, functional platform that ended up winning us 2nd place. A rollercoaster ride—worth every second!