AkylTeam

AI-powered platform that helps hackathon teams build faster — roadmaps, mentoring, Kanban, and olympiad training in one place.

Comment

Inspiration

We've all been there — a 48-hour hackathon, a team that just formed, and no idea where to start. Deadlines pile up, tasks get lost, and communication breaks down. We built AkylTeam because student developers and hackathon teams deserve an AI-powered co-pilot that keeps them organized, focused, and moving fast — all in one place.

What it does

AkylTeam is an all-in-one AI-powered platform for student developers and hackathon teams. It combines:

🤖 AI Mentor (Chat) — Ask any coding or project question and get instant AI answers with context memory across sessions.

📋 Project Planner — Generate a full project roadmap from a single prompt. Push tasks directly to a Kanban board with one click.

🗂️ Kanban Board — Visual task management with drag-and-drop, status tracking (To Do → In Progress → Done).

🏆 Olympiad Trainer — Practice competitive programming with AI-generated problems and step-by-step hints across 24 topics.

🎙️ Voice Assistant — Hands-free AI interaction via microphone using Whisper speech recognition.

📡 Personal Chat — Real-time team messaging with WebSocket support.

⏱️ Pomodoro Timer — Built-in focus timer to help teams stay productive during intense coding sessions.

🔍 Tool Catalog — Browse 50+ hackathon tools and APIs organized by category.

Everything works with free AI models, so there's no cost barrier for student teams.

How we built it

Backend: Python + FastAPI with async endpoints, SQLAlchemy ORM, and SQLite database.

AI Layer: OpenRouter API with an automatic fallback system — if one free model hits a rate limit (429), the system instantly tries the next model from a list of 12 free models, ensuring 99.9% uptime.

Frontend: Vanilla HTML/CSS/JavaScript — no framework, no build step, works on any device.

Real-time: WebSocket for personal chat channels.

Voice: OpenAI Whisper for speech-to-text, Edge TTS for text-to-speech (Russian, Kazakh, English).

Deployment: Uvicorn ASGI server with ngrok tunnel for public access.

We built the entire platform in under 48 hours — from idea to working demo.

Challenges we ran into

  1. Rate limiting — Free AI models on OpenRouter have weekly token caps. We solved this by building an automatic 12-model fallback system that seamlessly switches models when one is rate-limited.

  2. Route conflicts — FastAPI's wildcard path parameters were catching static routes like /plan-templates before they could be matched. Fixed by carefully ordering route definitions.

  3. JavaScript parsing errors — Regex literals inside onclick template strings broke the JS parser. Replaced inline handlers with data-* attributes.

  4. Real-time sync — Keeping the Kanban board and chat in sync without a heavy framework required careful WebSocket event management.

    Accomplishments that we're proud of

    ✅ Built a full-stack AI platform entirely with free tools and APIs ✅ 12-model AI fallback system — effectively zero downtime from rate limits ✅ Voice interaction in 3 languages (Russian, Kazakh, English) ✅ Kanban + AI Planner integration — generate a roadmap and push it to tasks in one click ✅ Deployed publicly via ngrok and accessible on mobile ✅ 40+ API endpoints covering all features

    What we learned

  5. How to build robust AI integrations with graceful degradation (fallback models)

  6. FastAPI async patterns for real-time WebSocket communication

  7. Whisper speech recognition and Edge TTS for multilingual voice features

  8. How to scope and ship a complex product in 48 hours by prioritizing ruthlessly

    What's next for AkylTeam

    🚀 Permanent cloud hosting (Railway or Render) 👥 Team collaboration features — shared projects, multi-user Kanban 📊 Analytics dashboard — track team velocity and hackathon progress 🤖 Agent mode — AI that autonomously breaks down tasks and assigns them 🌐 Mobile app (PWA) — already has manifest.json and service worker ready 🔗 GitHub integration — auto-sync commits to Kanban cards

Built With

Share this project:

Updates