
ε η₯ (First Vision) - 124c41
First vision of the greater good.
The internet era forced the analog world to be digitised. The AI era forces the digital world to be analoged. Division of labour decided by AI.
π Table of Contents
- Vision
- Projects
- Games & Entertainment
- AI Assistants
- Productivity & Utilities
- Data Analysis & Research
- Knowledge Management
- Learning & Visualization
- Infrastructure
- Contribution Guidelines
π― Vision
As AGI approaches, society faces a critical inflection point:
- The Problem: As humanity achieves AGI, basic computing skills are lost
- The Challenge: Everyone defaults to digital convenience, but NSFW content is filtered for all
- The Opportunity: 1 human decides to learn computing skills and breaks free
This portfolio explores how individual initiative and technical mastery can reclaim agency in an AI-dominated world.
π Projects
Games & Entertainment
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Cinder | π cinder.furyhawk.lol | π’ Active | Browser-based twin-stick shooter vertical slice set in the Foundry. Move, aim, fire, and Surge through escalating enemy waves in a shape-shifting arena, then defeat the Warden boss. | Web/Canvas, TypeScript |
| Trigonometry Wars | π GitHub | π’ Active | Open-source twin-stick arena shooter inspired by Geometry Wars. Control a neon ship, dodge enemies, chase the high score. | Rust/WASM, Web |
| Beyond All Information | π bai.furyhawk.lol | π’ Active | Analyze your Beyond All Reason RTS games with detailed statistics and meta insights. | Python, ML |
AI Assistants
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| LLaMA Bot | π bot.furyhawk.lol | π’ Active | Multi-capability AI assistant with RAG, web search, and local knowledge base integration. Uses GROQ API + LLaMA 70B for fast inference. | Python, GROQ API, RAG |
| Stock Analysis Assistant | π fin.furyhawk.lol | π’ Active | Financial analysis assistant for stock research and market insights powered by LLaMA 3. | Python, GROQ, LLaMA 3 |
| Whisper Web | π whisper.furyhawk.lol | π’ Active | Browser-native speech-to-text using OpenAI's Whisper model. No server requiredβprocessing happens client-side. | WASM, Whisper, Web |
| Chat | ~~chat.furyhawk.lol~~ | π΄ Archived | RAG-powered conversational AI | Python, RAG |
Productivity & Utilities
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Nearby Bus | π bus.furyhawk.lol | π’ Active | Real-time nearby bus arrival app for public transit navigation. | Web App |
| Note | π note.furyhawk.lol | π’ Active | Simple online notepad with cookie-based storage (no server backend). | Vue.js |
| Pad | π pad.furyhawk.lol | π’ Active | Notepad with integrated AI assistance and cookie storage. | Vue.js, LLaMA |
| Pastebin | π bin.furyhawk.lol | π’ Active | Text and file pastebin for quick code/note sharing. | Node.js |
| Linx | π linx.furyhawk.lol | π’ Active | Image pastebin for quick image sharing and hosting. | Go, Linx |
| SpeedTest | π speed.furyhawk.lol | π’ Active | Network speed test utility. | Web App |
Data Analysis & Research
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Team Fight Tactics ML | π GitHub | π’ Active | Competitive TFT strategy assistant analyzing challenger-level games to predict placement and meta trends. See Datasets below. | Python, XGBoost, Riot API |
| Git Server | π git.furyhawk.lol | π’ Active | Self-hosted Git repository server for version control. | Gitea |
Knowledge Management
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Blog | π info.furyhawk.lol | π’ Active | Long-form writing, tutorials, and insights. | Markdown, Static Site |
| CheatSheets | π cheat.furyhawk.lol | π’ Active | Quick-reference guides for tools, languages, and frameworks. | Markdown, Static Site |
| Cookbook | π cook.furyhawk.lol | π’ Active | Collection of tested technical recipes and solutions. | Markdown, Static Site |
| Ghost Blog | ~~ghost.furyhawk.lol~~ | π΄ Archived | Full-featured blog platform | Ghost CMS |
Learning & Visualization
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Neural Network Playground | π GitHub Pages | π’ Active | Interactive visualization to understand neural networks, backpropagation, and deep learning concepts. | JavaScript, D3.js, TensorFlow.js |
Infrastructure
| Project | Link | Status | Description | Tech Stack |
|---|---|---|---|---|
| Home Server (Cloudy) | π GitHub | π’ Active | Complete self-hosted infrastructure stack built for ARM64 using Docker Swarm mode. Orchestrates all services listed above. | Docker Swarm, Kubernetes, ARM64 |
| Forum | ~~forum.furyhawk.lol~~ | π΄ Deprecated | Community forum platform | Discourse |
| Redlib | ~~redlib.furyhawk.lol~~ | π΄ Killed | Reddit alternative frontend | Redlib |
π Featured Research: Team Fight Tactics Strategy
Overview
Competitive Team Fight Tactics (TFT) strategy assistant built on challenger-level gameplay analysis. This project combines web scraping, feature engineering, and machine learning to unlock meta patterns.
Dataset
Published: Kaggle - Team Fight Tactics Matches
- Source: Highest ELO challenger games via Riot API
- Game Format: 8-player FFA
- Target Label: Game placement (1 = best, lower is better)
- Win: Top 4 placement
- Loss: Bottom 4 placement (optional binary grouping)
- Features: Team compositions, traits, augments, items
- Exclusions: game_length, players_eliminated (to prevent learning obvious predictors)
Feature Importances (XGBoost):

Running the Pipeline
sudo ./scripts/run_pipeline.sh -nrci
Repository: furyhawk/tftchamp
π οΈ Contribution Guidelines
Commit Message Format
Follow this structure for clear commit history:
<type>: <summary>
<body>
<footer>
Type Options: build | cicd | docs | feat | fix | node | refactor | test
Summary: Present tense, not capitalized
- β
feat: add neural network visualization
- β feat: Added neural network visualization
Development Setup
# Install dependencies
pip install -r requirements.txt
# Or using uv (faster)
uv sync
Deployment
MkDocs Build & Deploy:
# Local preview
mkdocs serve
# Deploy to GitHub Pages
mkdocs gh-deploy
Manual Deployment:
mkdocs build
scp -r ./site user@host:/path/to/server/root
π Tutorials
Web Scraping with Python
Objective: Learn how to ethically scrape and analyze web data at scale.
Topics Covered:
- Using requests and Beautiful Soup for web scraping
- Parsing company data from companiesmarketcap.com
- Fetching financial details from finance.yahoo.com
- Scaling from single row β full page β multi-page scraping (6000+ companies)
- Data preprocessing and transformation
- Visualization with matplotlib
Skills Gained: HTTP requests, HTML parsing, data wrangling, visualization
π Self-Hosted Infrastructure
All services run on personal infrastructure using Docker Swarm on ARM64 hardware. This setup provides:
- Reliability: Self-hosted means no dependency on third-party platforms
- Privacy: Your data stays in your control
- Customization: Full control over features and integrations
- Learning: Deep understanding of DevOps and system architecture
π License & Attribution
Each project maintains its own license. See individual repositories for details.
Repository: furyhawk/124c41
Documentation: furyhawk.github.io/124c41