CareerForge: AI Agents That Interview You Back

College placement prep has a cold-start problem: students don't know where they actually stand, so they prepare for the wrong things. CareerForge attacks that head-on.

Step 1: Measure the student, not their claims

Instead of asking students to self-assess, CareerForge profiles them from the places their skills already leave evidence: GitHub, LinkedIn, LeetCode, HackerEarth and similar platforms. Pull the signal, score the skills, and place the student at an honest level.

Self-assessment says "I know DSA". Your LeetCode history says whether that's true.

Step 2: A roadmap that starts from reality

With the level established, the platform factors in what the student wants to become and how much time they have, then generates a personalized roadmap — with AI assistance at every step along the way, not just a static PDF of topics.

Step 3: The AI interview

The part I care most about: live AI agents conduct mock interviews — and they watch as well as listen. The interview pipeline runs facial and emotion detection during the session, so feedback covers not just what you answered but how you held up under pressure.

An interview you can retake infinitely, that never gets tired, and that grades you honestly — that's the practice loop most students never get access to.

Under the hood

CareerForge is built as a microservice architecture — profile ingestion, skill scoring, roadmap generation, and the interview agents are separate services, which is what lets each piece evolve independently (and lets the compute-heavy vision pipeline scale on its own).

It's the same thesis as everything else I build: measure reality first, then automate the feedback loop.