Build for a Friend: An Interactive DSA and Algorithm Coach for Competitive Exam Prep

What I Built

I built the GATE CS Visual Dry-Run Tracer for a close friend who is aggressively preparing for highly competitive core engineering exams like GATE CS, ISRO, and DRDO.

They kept losing critical marks on technical algorithm evaluation papers because standard compilers only print out a final code output—they don't show you where your manual dry-run logic diverged at step 3 or 4 of a 10-step execution trace.

This app breaks down any complex code block, nested loop, or graph algorithm problem into an interactive step-by-step matrix table. It allows the student to commit their intermediate checkpoint values in a scratchpad before revealing the absolute step-by-step truth, making it a perfect tool for diagnosing logical slips under exam time pressure.

Demo

👉 Launch the GATE CS Visual Dry-Run Tracer App

Code

🔗 View the Application Source Repository

How I Built It

The application is built completely on top of an open-source architecture:

• The Core Brain: We utilize the open-weights GPT-OSS-20b model, processed through a high-speed open-access Groq API client pipeline.
• Interface & UX: Designed a clean, distraction-free grid-split interface using the open-source Gradio framework.
• Deployment Infrastructure: Deployed as a standalone application inside Hugging Face Spaces utilizing the open-access ZeroGPU compute tier wrapper.
• Prompt Harness: Engineered custom low-temperature (temperature=0.2) instruction sets to enforce strict, predictable mathematical markdown transformations from natural text input into deterministic tables.

Why Does Open Innovation Matter?

Open innovation was the only reason this project was possible to ship over a weekend without a corporate budget:

1. Infinite Execution Token Capacity: Tracing complex algorithm loops step-by-step or detailing matrix transitions consumes thousands of context tokens per query. Standard commercial pay-per-token APIs would bankrupt a student using it for daily prep. Open weights on open-access endpoints mean infinite free dry-runs.
2. Deterministic Stability: Proprietary models frequently drift in behavior overnight due to black-box alignment updates. By grounding our tool in an open-weights standard with zero temperature variations, we enforce static academic formatting required for error-free textbook analysis.

Prize Categories

• Open Source AI Novelty
• Build for a Friend