AI Placement Predictor: Employability DNA Engine

Explainable AI that predicts student placement readiness, shows exactly why, and builds a personalized upskilling roadmap, while giving TPOs cohort analytics to train at-risk students early.

Comment

Inspiration

Every year, placement offices struggle to answer one question: which students are truly placement-ready, and who needs help? Most students only find out after being rejected, with no idea why. Placement readiness depends on much more than CGPA: certifications, projects, internships, aptitude, communication and extracurriculars all matter. We wanted to build a system that gives students clear, explainable answers and gives Training & Placement Officers (TPOs) data to act early.

What it does

AI Placement Predictor: Employability DNA Engine is an internal institutional platform for students and TPOs.

  • Student Profile & Data Ingestion: Students enter their academic, technical, project, aptitude, soft-skill and extracurricular details manually or by CSV upload.
  • ML Placement Probability: A trained model gives a 0-100% placement probability and a readiness status: Ready, Near-Ready or Needs Training.
  • Career Track Mapping: Predicts alignment and confidence for tracks like Full-Stack Developer, Data Analyst, Cloud/DevOps Engineer and QA Specialist.
  • Explainable AI: SHAP/LIME visualizations show which factors boosted or reduced the score (e.g., "+18% due to full-stack internship, -14% due to missing database certification").
  • Skill Gap Diagnosis & Roadmap: Compares the student's skills with career-track benchmarks and generates a phased upskilling plan with courses, project ideas, interview drills and time estimates.
  • TPO Dashboard: Shows department-wise readiness, flags students below 60% readiness for targeted bootcamps, and displays skill-deficit heatmaps (e.g., "64% of 6th-semester students lack SQL").

How we built it

  • Frontend: React.js with Tailwind CSS and Plotly/Recharts for interactive dashboards (or Streamlit for rapid prototyping)
  • Backend: FastAPI REST APIs for prediction, explanation, skill-gap, roadmap and TPO analytics
  • ML Pipeline: Python, Pandas, NumPy and Scikit-learn for preprocessing and evaluation, with XGBoost / Random Forest as the main model and Logistic Regression as a baseline
  • Explainability: SHAP and LIME for per-student feature contributions
  • Skill-Gap Engine: Benchmark-based comparison against career-track profiles to generate roadmaps
  • Data: Anonymized/synthetic student dataset plus CSV upload; SQLite for storage
  • Tools: GitHub, Google Colab, Postman

The system runs entirely inside the college ecosystem, with no external job portals or scraping.

Challenges we ran into

  • Building a realistic synthetic dataset that reflects real placement patterns without using private student data
  • Keeping the model accurate while still making its predictions easy to explain
  • Turning SHAP outputs into simple, student-friendly insights instead of raw numbers
  • Designing role benchmarks and roadmaps that are realistic and time-bound
  • Fitting the full pipeline, from data to model to explanation to dashboard, into a short time frame

Accomplishments that we're proud of

  • A complete end-to-end design covering prediction, explanation, diagnosis and institutional analytics in one platform
  • An explainability-first approach, so no student gets a score without knowing why
  • A dual-purpose design: it helps individual students improve and helps TPOs plan department-wide training
  • A privacy-conscious architecture that works on anonymized data and keeps everything inside the institution

What we learned

  • Explainability builds trust: students act on predictions they can understand
  • Employability is multidimensional, and academics alone are a weak predictor
  • Good ML work depends heavily on clean data, thoughtful feature engineering and honest evaluation
  • Institutional tools need both an individual view for students and an aggregate view for administrators

What's next for AI Placement Predictor: Employability DNA Engine

  • Train on real anonymized historical placement data from the institution
  • Add more career tracks and company-specific readiness profiles
  • Integrate mock interview and aptitude test modules for live skill assessment
  • Add an LLM-powered mentor to personalize roadmap guidance
  • Track student progress over semesters to show improvement trends
  • Add role-based login, notifications for at-risk students, and multi-college support

Built With

Share this project:

Updates

Submission history