aka champion19007 github profile

📬 About me — Contact

My work sits at the intersection of LLMs, Computer Vision, and Reinforcement Learning —
I enjoy designing agents that can reason, plan, see, and act with autonomy.
Whether it's training models, crafting datasets, building pipelines, or deploying systems,
I care about engineering that is clean, scalable, and grounded in real research.

I’ve worked across many parts of the AI stack:
from transformer models and RAG pipelines to ROS2 robotics systems,
from CV model training to hybrid RL + search algorithms.
I’m constantly exploring how different branches of AI can be combined
to create something more powerful than their parts.

🛠️ Skills

Programming

Frameworks & Libraries

Tools & Platforms

Domains

---

🔧 What I Do Now Build LLM Agent Systems Architect RAG & Retrieval Pipelines Train & deploy Computer Vision Models (CNNs, ViTs, YOLO) Develop Reinforcement Learning + Search Engineer Backend Architectures (FastAPI, Python, PostgreSQL) Research hybrid engines (Search + RL + Transformers)

✨ I Can Help With LLM Agents — planning, tool use, memory, knowledge graphs Computer Vision — detection, segmentation, transformers Reinforcement Learning — hybrid MCTS models, game-AI, reward engineering Backend for AI Inference — scalable model-serving systems Optimizing PyTorch Pipelines — speed, memory, performance

🌟 Featured Projects

♟️ Hybrid Minimax–MCTS Chess Engine A patent-inspired hybrid chess engine merging Minimax, MCTS, and deep evaluation networks.	🤖 RL Chess Engine — Deep RL + Search A Reinforcement Learning chess engine combining policy/value networks with Monte Carlo Tree Search.
📈Advanced Stock Forecaster — CNN-BiLSTM-GRU A hybrid CNN + BiLSTM + GRU deep-learning model for time-series trend prediction, capturing temporal patterns and long-range dependencies for improved stock movement forecasting.	🐾 Real-Time Animal Detection (CCTV Vision) A computer-vision pipeline using YOLO + OpenCV to detect animals in real time from CCTV streams, built for safety monitoring and rapid alerting.

📝 Docs Workflow

I use a unified research workflow that connects notes → videos → documentation → experiments → implementation.
Designed for clarity, speed, and high-signal learning.

📚 Notes & Knowledge Base Obsidian — Main AI/ML knowledge base https://obsidian.md Markdown — Universal format for logs, drafts, and papers Notion — Structured documentation + planning https://www.notion.so Excalidraw — ML diagrams, architecture sketches https://excalidraw.com Draw.io (diagrams.net) — System & pipeline diagrams https://app.diagrams.net 🎥 Videos & Learning Channels Andrej Karpathy — Transformers, GPT, LLM systems https://www.youtube.com/@AndrejKarpathy Two Minute Papers — Fast research summaries https://www.youtube.com/@TwoMinutePapers Yannic Kilcher — Deep dives on new papers https://www.youtube.com/@YannicKilcher Sentdex — Python, ML, CV tutorials https://www.youtube.com/@sentdex DeepMind Talks — High-quality ML research content https://www.youtube.com/@deepmind Microsoft Research — LLM systems & agent research https://www.youtube.com/@MicrosoftResearch

📄 Official Documentation / References PyTorch Docs — https://pytorch.org/docs Transformers Docs — https://huggingface.co/docs/transformers OpenAI API Docs — https://platform.openai.com/docs LangChain Docs — https://python.langchain.com/docs Ultralytics YOLO — https://docs.ultralytics.com Gymnasium (RL) — https://gymnasium.farama.org Stable Baselines3 — https://stable-baselines3.readthedocs.io 🧪 Experimentation & Tracking Jupyter Notebook — https://jupyter.org TensorBoard — https://www.tensorflow.org/tensorboard Weights & Biases — https://wandb.ai Git — versioned experiments, reproducible research https://git-scm.com 🔧 Useful Tools OpenAI Playground — https://platform.openai.com/playground HuggingFace Inference API — https://huggingface.co/inference-api Papers With Code — https://paperswithcode.com arXiv — https://arxiv.org Ollama — https://ollama.com LM Studio — https://lmstudio.ai

Notes → Documentation → Experiments → Deployment Everything stays tightly integrated and optimized for deep-work.

⭐ Recent activity

Click to expand recent GitHub activity

🔁 Fresh Pull Requests

• 🟣 feat: as requested in #48: sort icons according to window position in hyprland-community/hyprland-autoname-workspaces • 4 weeks ago
  _{Hyprland autoname workspaces 🪟 [maintainers=@cyrinux,@maximbaz,@matt-fff]}

• 🟢 Restrict Type/Class scope to user-defined symbols in aaronedev/violet-void-theme_subl • 1 month ago

• 🟢 Rename theme to Violet Void in aaronedev/violet-void-theme_subl • 1 month ago

• 🟢 Fix script directory resolution for symlinked runs in aaronedev/nerdfont-icon-rofi-picker • 1 month ago
  _{A simple rofi-based picker for nerd font icons. Browse through thousands of icons with fuzzy search and copy them to your clipboard.}

• 🟢 Preserve locale variables when launching rofi in aaronedev/nerdfont-icon-rofi-picker • 1 month ago
  _{A simple rofi-based picker for nerd font icons. Browse through thousands of icons with fuzzy search and copy them to your clipboard.}

✨ WakaTime stats ✨

Click to expand the latest metrics

I'm an Early 🐤

🌞 Morning                54 commits          ⬛⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   22.88 % 
🌆 Daytime                97 commits          ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   41.10 % 
🌃 Evening                46 commits          ⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   19.49 % 
🌙 Night                  39 commits          ⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   16.53 % 

📅 I'm Most Productive on Monday

Monday                   67 commits          ⬛⬛⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   28.39 % 
Tuesday                  15 commits          ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   06.36 % 
Wednesday                37 commits          ⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   15.68 % 
Thursday                 10 commits          ⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   04.24 % 
Friday                   40 commits          ⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   16.95 % 
Saturday                 18 commits          ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   07.63 % 
Sunday                   49 commits          ⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   20.76 % 

📊 This Week I Spent My Time On

🕑︎ Time Zone: Europe/Berlin

💬 Programming Languages: 
Lua                      7 hrs 25 mins       ⬛⬛⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   28.42 % 
Markdown                 5 hrs 17 mins       ⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   20.26 % 
tmux                     2 hrs 22 mins       ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   09.07 % 
Bash                     2 hrs 8 mins        ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   08.19 % 
hyprlang                 1 hr 35 mins        ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   06.07 % 

🔥 Editors: 
Neovim                   24 hrs 37 mins      ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜   94.32 % 
Vim                      1 hr 5 mins         ⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   04.16 % 
Cursor                   22 mins             ⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   01.42 % 
Android Studio           1 min               ⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   00.10 % 

🐱‍💻 Projects: 
dotfiles                 19 hrs              ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜   72.78 % 
Unknown Project          3 hrs 40 mins       ⬛⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   14.09 % 
cheatsheets              2 hrs 27 mins       ⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   09.44 % 
Adwaita-for-Steam        17 mins             ⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   01.11 % 
wezterm_resurrect_dude   14 mins             ⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   00.91 % 

💻 Operating System: 
Linux                    26 hrs 6 mins       ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛   100.00 % 

I Mostly Code in Shell

CSS                      11 repos            ⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   12.50 % 
Lua                      10 repos            ⬛⬛⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   11.36 % 
Dockerfile               5 repos             ⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   05.68 % 
JavaScript               5 repos             ⬛⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   05.68 % 
TypeScript               1 repo              ⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜⬜   01.14 %