AI Placement Coach

AI Placement Coach is an realistic, AI-powered interview practice platform delivering company-specific coding and CS theory evaluations with structured feedback to simulate real placement interviews.

Comment

Inspiration

As a student preparing for my internship and placement season, I practiced a lot of problems on platforms like LeetCode and Codeforces, studied core CS subjects, and went through many questions on GeeksforGeeks and similar sites.

However, even after doing all this, I was never fully confident about how I would actually perform in a real interview — how I would explain my approach, answer theory questions clearly, or think under pressure. Most platforms helped with practice, but none of them felt like a real interview.

This gap between solving problems and performing in interviews inspired us to build AI Placement Coach.

What it does

AI Placement Coach is an AI-powered interview practice platform that simulates real placement interviews across multiple subjects:

  • DSA (coding interviews)
  • Operating Systems
  • DBMS
  • OOPS

For DSA, users solve company-specific coding problems and optionally explain their approach using voice.
For theory subjects, users answer interview-style questions in timed, session-based flows.

The platform evaluates answers using AI and provides structured feedback, including scores, strengths, improvement areas, and missed concepts — similar to how an interviewer would assess a candidate.

How we built it

We built the frontend using React + Vite, with Monaco Editor to provide a professional coding environment similar to real interviews.

The backend is built with Node.js and Express, deployed on Google Cloud Run.
We used Google Vertex AI (Gemini 2.5 Flash) to evaluate answers, Text Embeddings for semantic and hybrid search, and Cloud Speech-to-Text to transcribe spoken explanations.

The system uses curated question banks, session tracking (to avoid repetition), and structured evaluation prompts to ensure consistent feedback.

Challenges we ran into

One of the main challenges was designing consistent AI evaluation for open-ended answers. Different users explain concepts differently, so we had to use strict rubrics and carefully designed prompts to reduce variability.

Handling multimodal inputs (code + explanation audio), managing session-based question rotation, and deploying a full-stack application on cloud infrastructure within limited time were also significant challenges.

Accomplishments that we're proud of

  • Built a fully working MVP with live deployment
  • Implemented multimodal evaluation (code + explanation)
  • Designed an interview-like user experience instead of a simple practice app
  • Integrated multiple Google Cloud services into a production-ready system
  • Delivered meaningful, structured feedback beyond just right or wrong answers

What we learned

Through this project, we learned how to:

  • Design AI evaluation systems using structured prompts
  • Work with vector embeddings and semantic search
  • Deploy scalable applications using Google Cloud Run
  • Build AI systems that balance intelligence with reliability
  • Think from an interviewer’s perspective, not just a student’s

What's next for AI Placement Coach

In the future, we plan to add:

  • User authentication and personalized profiles
  • A progress dashboard to track improvement over time
  • A mock interview mode with timed, end-to-end interview sessions
  • Adaptive difficulty and deeper personalization based on performance trends

Our long-term goal is to make AI Placement Coach feel as close as possible to a real technical interview experience.

Built With

Share this project:

Updates