Place Path — AI Placement Assistant

Inspiration

College students often practice many questions but still struggle to understand what they are actually weak at. Generic recommendations usually react to a single wrong answer instead of looking at persistent patterns.

We wanted to build a placement coach that identifies genuine skill gaps from repeated assessment behavior and then turns those gaps into a personalized learning path.

What we built

Place Path is an AI-powered placement assistant for college students.

Instead of treating one wrong answer as a skill gap, Place Path analyzes repeated performance patterns, question difficulty, consistency, and error behavior to identify persistent weaknesses.

The flow is:

Assessment → Multi-attempt Analysis → Skill Gap Detection → Confidence Check → Targeted Diagnostic → Personalized Learning Path → Reassessment

This creates a continuous assess → diagnose → learn → reassess loop.

How it works

A student's assessment attempts and performance data are analyzed to build a skill profile.

When the evidence is weak, the system can use a targeted diagnostic question before making a stronger skill-gap conclusion.

The recommendation layer then produces a personalized learning path focused on the student's weak areas.

Why it is different

A simple approach is:

Wrong answer → Recommend a similar question

Place Path aims for:

Multiple attempts → Error pattern analysis → Skill-gap detection → Confidence estimation → Targeted diagnostic → Personalized recommendation

Technical approach

The prototype uses a React/TypeScript frontend, a Node.js backend with REST APIs, PostgreSQL for data, and AI services for analysis and recommendations. The AI architecture is designed around structured outputs and validation before results are exposed to the frontend.

Challenges

The main challenges were sparse student history, avoiding false skill-gap diagnoses, maintaining consistent AI outputs, and keeping recommendations useful.

We addressed these with minimum evidence thresholds, confidence-aware analysis, rule-based validation, and controlled test cases.

What we learned

We learned that useful AI placement guidance needs more than a single assessment score. Looking at performance patterns over multiple attempts gives a better basis for deciding whether a weakness is persistent and what the student should learn next.

Future direction

The prototype is designed to grow into a broader placement-preparation platform with continuous skill tracking, targeted practice, personalized roadmaps, mock interviews, progress tracking, and premium capabilities.

Built With

• ai
• api
• css
• express.js
• llm
• node.js
• postgresql
• react
• redis
• rest
• revenuecat
• tailwind
• typescript
• vite