Coding Flora

Overloaded by thousands of LeetCode questions? Master the underlying patterns. Solve one problem and master a hundred! Just 5 minutes a day, land your dream tech job within a year.

Comment

Inspiration

Every year, 1 million people apply to Google—but less than 0.2% get in. We discovered that when students practice coding problems, they waste up to 70% of their time on wrong attempts—over 400–600 hours lost. For underserved communities, that’s not just time wasted, it’s opportunity lost.

What it does

Coding Flora is a voice-powered AI coding coach built on Amazon Bedrock AgentCore. Learners pick up their phone, speak their ideas out loud, and get instant feedback—6× faster than typing. It enables visually impaired and time-constrained learners to practice coding in short 5-minute sessions, anytime, anywhere.

How we built it

We built an autonomous AI Agent using Amazon Bedrock AgentCore and Claude Reasoning models, giving the coach reasoning and planning abilities.

This setup lets the AI coach autonomously plan, check, and respond to learners—meeting AWS’s agent qualification standards.

Challenges we ran into

Designing prompt chains that capture students’ spoken logic instead of syntax

Integrating voice recognition with cloud to keep inference latency under 1 second

Maintaining context for multi-turn sessions between agent and LLM

Accomplishments that we're proud of

🥈 Won $10 000 and advanced the University of Chicago team to the World Championship at the 2024 AI for Science Hackathon

Supported 200–300 students in landing jobs at Google and Amazon

Projected to generate $20,000 annual revenue from university clients

Received a $100 K institutional offer to deploy our AI system campus-wide

What we learned

Amazon Bedrock AgentCore + Claude LLMs enable rapid, low-ops deployment of reasoning agents

Voice AI can democratize education access for students with limited time or visual impairments

Emotional feedback and inclusive design improve learning retention as much as technical accuracy

What's next for Coding Flora

DynamoDB + S3 store learning progress, transcripts, and daily streak data.

AWS Lambda + API Gateway process voice input, run evaluation scripts, and feed results back to Claude.

Built With

Share this project:

Updates