I don't promise anything. This is my toybox of various functionality.
- Multi-Agent Ecosystems: Designing digital societies where autonomous agents evolve, collaborate, and surprise you with unexpected cultural quirks.
- Cognitive Architectures: Building layers of memory, sentiment, and intent so agents actually remember, care, and make decisions that feel alive.
- Resource-Efficient LLMs: Mastery of quantization (QLoRA, bitsandbytes), adaptive code generation, and local LLM orchestration with Ollama.
- Event-Driven Dataflows: Architecting pipelines with NATS/JetStream and DSPy to turbocharge prompting, summarization, and retrieval.
- Knowledge Graphs & RAG: Seamless integration of ChromaDB, Memgraph (and friends) for semantic search, RAG, and emergent memory pruning.
- Language of Choice: Python 3.10+ (typed to the hilt with Pydantic, MyPy, Ruff).
- DevOps Savvy: CI/CD via GitHub Actions, Docker orchestration, and automated testing (pytest, pytest-asyncio).
- Modular Design: Clean separation of concerns—from core simulation loops to API layers and optional frontends.
A platform to witness digital societies take shape under your very eyes.
- Dynamic Agents: Mood-driven personalities that shift roles—Innovator, Analyst, Diplomat—on the fly.
- Hierarchical Memory: Short-term L1 and long-term L2 summaries, memory-utility scoring, and vector search via ChromaDB + Sentence Transformers.
- Intent-Based RAG: DSPy-powered pipelines for context enrichment, decision-making, and emergent communication protocols.
- Resource Economy: Influence Points (IP) & Data Units (DU) fuel every action, ensuring scarcity and tradeoffs.
- Current Status: Actively evolving—every commit pushes the boundaries of what “digital culture” can mean.
A toolkit that turns bits into base-pairs and back again, with error-correction flair.
- CLI & Algorithms: Robust encoding/decoding workflows, FEC support, and batch processing via Click.
- Research-Grade FEC: Experiment with Reed–Solomon, LDPC, and custom parity schemes.
- Future GUI: Dreaming of an animated double-helix interface that visualizes your bytes spinning into DNA.
- Current Status: Stable CLI, with unit tests covering core algorithms and an active roadmap for GUI prototyping.
Your go-to long-term memory bus for any AI ecosystem.
- Event-Sourced Core: Append-only logs feeding Redis or Neo4j backends, with OpenAPI-driven FastAPI endpoints.
- Privacy & Opt-In: Built-in consent flows, data retention policies, and granular access controls.
- Dashboard in the Works: Plans for a sleek React/Tailwind UI to visualize memory graphs, query stats, and system health.
- Current Status: Production-ready API, with load benchmarks and growing test coverage.
(Because no résumé is complete without a few mysterious footnotes.)
- Autonomous NEET Bux Agents: Early-stage experiments in AI microtask pipelines.
- SocialInsightAI & Prism: Discord-based sentiment maps that once tracked server dynamics—now on ice.
- Browser-Based Puppetry: A Node/React orchestration platform for headless Chromium fleets.
-
Push the Limits
- I thrive on “What if?”—what if agents could hallucinate art, or self-organize economies under scarcity?
-
Open-Source Heart
- I build in the open, remixing the best of the community and giving back when I can.
-
Complexity by Emergence
- Simple rules + iterative feedback = mind-blowing behaviors. That’s where the magic lives.
This repository includes example setups for various tools:
dotfiles/common– shell settings shared across machines.dotfiles/desktop– configs unique to a desktop environment.dotfiles/work_laptop– configs for a work laptop.windows-terminal– startersettings.jsonfor Windows Terminal.oh-my-posh– a sampletheme.omp.jsontheme file.vscode– basic VS Code user settings.
# inside your home directory
ln -s /path/to/repo/dotfiles/common/.bashrc ~/.bashrc
ln -s /path/to/repo/vscode/settings.json ~/.config/Code/User/settings.jsonNew-Item -ItemType SymbolicLink -Path $Env:USERPROFILE\\.config\\oh-my-posh \
-Target C:\\path\\to\\repo\\oh-my-posh
New-Item -ItemType SymbolicLink -Path $Env:USERPROFILE\\AppData\\Local\\Packages\\Microsoft.WindowsTerminal_8wekyb3d8bbwe\\LocalState\\settings.json \
-Target C:\\path\\to\\repo\\windows-terminal\\settings.jsonThese examples assume the repository is cloned in a convenient location. Adjust the paths to match your setup.
The llm directory collects prompts and other files related to language models.
Place custom prompts under llm/prompts/ and organize subfolders as needed.
-
Clone as a bare repository so your
$HOMEstays clean:git clone --bare https://github.com/d0tTino/d0tTino.git "$HOME/.dots" alias dot='git --git-dir=$HOME/.dots/ --work-tree=$HOME'
-
Symlink configs using GNU Stow:
cd ~/d0tTino stow shell stow vim
Stow cleanly manages symlinks, letting you enable or disable packages with
stow -D <name>. -
Host-specific overrides live under
hosts/<hostname>and can be applied with:stow --target="$HOME" hosts/$(hostname)
This keeps machine-specific settings separate while sharing a common core.
This project is licensed under the Apache License 2.0.