interSPED

• 

  Dashboard

• 

  Landing Page

• 

  Before joining the call

• 

  Interview Platform

InterSPED is an AI-powered mock interview platform that leverages a hybrid technology stack to simulate realistic interview experiences and provide actionable feedback.

Inspiration

With many students actively seeking summer internships and jobs, interview preparation has become increasingly important. We identified a need for a platform that allows users to practice interviews in a realistic, low-pressure environment tailored to specific roles and companies.

What It Does

Users input a job posting and a target company. InterSPED then conducts a mock interview customized to that role, simulating real interview questions and interactions to help users prepare effectively.

How We Built It

Back End: FastAPI (Python) and C++

Front End: React / Next.js

Supporting Technologies: WebSockets, Docker, OpenCV, Presage SDK

Challenges We Ran Into

Presage SDK limitations Presage was unavailable for Python and unsupported on Windows. Solution: We installed and configured WSL to build and run the C++ implementation in a Linux environment.

Library version incompatibilities Presage’s C++ installation required a newer library version than what was available. Solution: With guidance from a staff member, we modified the installation process to support a lower-compatible version.

Real-time camera incompatibility with WSL OpenCV’s real-time camera access does not work reliably in WSL. Solution: We adapted the pipeline to record video locally and process recordings instead of live streams.

C++ and Python integration challenges Presage’s C++ implementation did not integrate cleanly with our Python-based backend. Solution: We used JSON files as an intermediary communication layer between the two systems.

Resource inefficiency Our early development model did not adequately account for API usage, leading to excessive API key consumption.

Accomplishments We’re Proud Of

Building a hybrid backend combining Python and C++, a design approach none of us had previously used.

Successfully running a Linux-only SDK on Windows via WSL.

Implementing real-time speech-to-text transcription using WebSockets for the first time.

Designing and delivering a polished, intuitive UI/UX on the frontend.

Using Docker to encapsulate and integrate Linux-based APIs into a Windows-compatible workflow.

What We Learned

How to design and deploy an agentic model using Solace.

The importance of containerization (Docker) for cross-platform compatibility and complex system integration.

What’s Next for InterSPED

Use Docker-based camera support to enable real-time vital analysis in addition to post-interview feedback.

Incorporate interviewer emotional modeling.

Improve sentiment and behavioral analysis.

Add a job scouting feature.

Implement a resume optimization tool.

Develop a personalized learning agent to guide long-term interview improvement.

Built With

• c++
• digitalocean
• elevenlabs
• github
• presage
• python
• react
• solace
• wsl