**This is a submission for the Hacktoberfest Open-Source AI Challenge — Week 2: Side Hustle.
My Orbits: Turn Your GitHub Projects Into a Tech Career Roadmap
What I Built
Starting a career in tech can be overwhelming. There are endless tutorials, technologies, and career paths, but beginners often struggle to figure out what they already know, which skills they're missing, and what to learn next.
My Orbits is an AI-powered career intelligence platform that turns a public GitHub repository into an interactive dashboard for understanding skills and planning a tech career.
It includes:
- Interactive 3D skill graph: Explore the technologies in a project and how they connect.
- Career readiness: Get a score against a selected career role.
- Skill gap analysis: Identify the skills worth prioritizing.
- Learning roadmaps: Follow a beginner-friendly plan with learning resources.
- What-If simulator: Explore how learning a new skill could affect your readiness.
- Market insights: Understand technology demand using available role benchmarks.
- Recruiter profile: Generate a shareable profile and export a PDF.
- AI Copilot: Ask questions and get recommendations grounded in your project's skill graph.
The idea is simple: paste a repository, choose a target role, and understand what you know, what's missing, and what to do next.
Who would pay for it?
I'm exploring two potential customer groups:
- Job seekers and career-switchers: A paid skill report could help them present project-backed skills and identify gaps before applying for jobs. An initial price to test is $7–$9 per report.
- Bootcamps and career coaches: A cohort dashboard could help track students' demonstrated skills and learning gaps before placement season. An initial pricing hypothesis is $29–$49 per month per student.
These are pricing hypotheses, not validated sales. My next step is to test whether students and career coaches find the reports valuable enough to pay for them.
Demo
Video demo: [https://vimeo.com/1237044249?share=copy&fl=sv&fe=ci]
Video demo: [https://youtu.be/fp53yd4m61M]
Source code: github.com/riishcodes/Myorbits
There isn't a hosted demo yet. You can run the project locally and explore the demo profile without signing up or providing an API key for the seeded demo flow.
Run locally
Backend
cd backend
python -m venv .venv
# Windows:
.venv\Scripts\activate
# macOS/Linux:
# source .venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8000
Frontend — in a second terminal:
cd frontend
npm install
npm run dev
Open http://localhost:3000 and select Try Demo.
The seeded demo profile contains 23 skills, 24 nodes, and 50 connections for a Full Stack Developer target. Its readiness score is 67.9%, with Node.js and SQL highlighted as gaps. In the demo, adding Node.js through the What-If simulator raises the score to 75.9%.
Note: Market figures are role benchmarks, not a live feed of job postings. Readiness scores are estimates, not guarantees of employability.
Code
- GitHub: https://github.com/riishcodes/Myorbits
- License: MIT
Key implementation areas:
-
backend/services/llm_service.py— shared AI provider integration -
backend/services/ai_service.py— skill extraction and analysis -
backend/services/chat_service.py— project-context Copilot -
backend/services/scoring_service.py— readiness scoring and What-If logic -
frontend/src/components/graph/— interactive 3D knowledge graph -
frontend/src/components/panels/— insights, roadmap, market, recruiter, and Copilot panels
My Orbits — Interactive 3D Knowledge Graph Builder
AI-powered Career Intelligence Platform that transforms GitHub activity, Notion notes, and certifications into measurable hiring intelligence using AI reasoning and interactive 3D knowledge graphs.
✨ Features
- 🔮 Interactive 3D Knowledge Graph — Force-directed, zoomable, clickable skill visualization
- 🔗 GitHub Integration — OAuth login, live repo fetching, language extraction
- 🤖 AI Skill Extraction — Open-weight LLM skill extraction (free) via OpenRouter or local Ollama
- 📊 Career Readiness Score — Quantified 0-100 with breakdown
- 🎯 Skill Gap Analysis — AI-powered gap identification with priority scores
- 🔮 What-If Simulator — "If you learn PyTorch, readiness +16%"
- 👔 Recruiter Mode — Radar chart + PDF export
- 📈 Analytics Dashboard — Pie, radar, area, and bar charts
- 🔍 Search & Filter — Real-time node search + category filter
- 🌊 Market Demand Overlay — Node glow intensity by demand
🏗️ Tech Stack
| Layer | Technology |
|---|---|
| Frontend | Next.js 14, TypeScript, |
How I Built It
My Orbits uses a Next.js 14, TypeScript, and Tailwind CSS frontend with a FastAPI backend. The graph uses 3d-force-graph and Three.js, Zustand manages state, Recharts renders analytics, and Framer Motion handles UI animations.
Open-weight AI at the core
The project supports two inference paths:
-
Hosted inference: Google Gemma 4 26B A4B IT through OpenRouter, using
google/gemma-4-26b-a4b-it:free. - Local inference: Qwen2.5 7B Instruct through Ollama, allowing model inference to run on the user's machine.
The AI_PROVIDER configuration selects the provider, and model IDs can be configured through environment variables. This makes it easier to switch models without rewriting the application.
The models support five key workflows:
| Feature | AI contribution |
|---|---|
| Skill extraction | Helps interpret repository evidence and map technologies to skills |
| Gap analysis | Explains missing skills in the context of a target role |
| Learning roadmap | Turns skill gaps into an ordered learning plan |
| What-If simulator | Explains potential changes when a skill is added |
| Copilot | Answers questions using the user's project and skill graph as context |
Not everything is left to the model. Repository parsing, the skill directory, and the readiness calculation provide a structured foundation for the results.
The readiness score uses a weighted formula: 40% skill coverage, 30% average maturity, 20% market alignment, and 10% project depth. The model helps explain the results rather than inventing the score.
Keeping the demo accessible
The seeded demo profile makes it possible to explore the dashboard without running a full repository analysis on every visit. The application can also analyze a public repository through its normal workflow.
The goal is to keep experimentation accessible while making the core AI capabilities replaceable and configurable.
Why Does Open Innovation Matter?
1. Local inference and greater control over data
Career profiles and repository contents can be sensitive. With Ollama configured for local inference, model processing can happen on the user's machine instead of sending that information to a hosted model provider. Fetching a new public repository still requires an internet connection, and other external services may have their own data flows.
2. Lower barriers for students
Students and career-switchers may not have the budget for paid AI APIs. Local models provide a way to experiment without per-token inference charges, while free hosted endpoints can make setup easier for users without suitable hardware. Both options have practical limits, including hardware requirements and hosted usage quotas.
3. Flexibility without vendor lock-in
A shared provider interface makes it easier to switch between compatible open-weight models. Developers can experiment with different models, run inference locally, or use hosted infrastructure without rebuilding the entire product around a single provider.
Open innovation makes My Orbits more adaptable and gives users more control over how its AI capabilities are delivered.
My Agent Session
I didn't capture a DevRelay session for this build, so I won't claim one. I used Claude Code as an AI pair-programming assistant during development.
Prize Categories
- Best Use of Gemma: My Orbits integrates Gemma 4 26B A4B IT through OpenRouter for its AI workflows, alongside a local Qwen2.5 option. This category is relevant if the challenge accepts hosted inference using an open-weight Gemma model.
- Overall challenge: Submitted for consideration in the overall challenge.
Tags: devchallenge hf26challenge ai opensource hacktoberfest**
Top comments (0)