I thought it would be nice to have a structured summary of my recent projects, so I had Claude build one from repos and jupyter notebooks on my laptop. It did a pretty good job - an auto generated list of projects and theme grouping is below:
Themes & Tools Used
Core Themes
| Theme | Description |
|---|---|
| LLMs from scratch | Building and understanding transformer architectures (nanoGPT, Llama 3.2 E2E, custom ALMOD, lllib) |
| AI agents & orchestration | Evolving from simple tool-calling → agent loops → multi-agent orchestration → kanban-driven autonomous agents |
| Event data platform | Long-running project: NYC event discovery, ETL, deduplication, and web UI (omnivore → event_etl → event_etl_revisited) |
| Speech & multimodal AI | TTS (Bark, XTTS, Kokoro, SNAC), speech LMs (Spirit LM, Mini-Omni) |
| Reinforcement learning | Stable Baselines 3, PPO from scratch, DQN, Godot RL agents |
| Quantitative finance | FRED/IBKR/DataBento APIs, GARCH volatility, pairs trading, prediction markets |
| Game dev + AI | Godot RPG with LLM-generated content, Gemini-generated HTML games, procedural narrative engines |
| Coding agent tooling | Claude Code templates, pi agent ecosystem, skill systems, iterative dev loops |
| Probabilistic programming | PyMC, Pyro, Bayesian methods |
| Business/consulting | AAO organization management, solar projects, staffing, exchange platforms |
Languages & Frameworks
| Category | Tools |
|---|---|
| Python | PyTorch, transformers, LangGraph, Guidance, Outlines, Stable Baselines 3, LightGBM, DuckDB, dbt, Dagster, dlt, FastAPI, Flask, Modal, gspread, SymPy, PyMC, Pyro |
| JavaScript/TypeScript | Svelte/SvelteKit, React, Astro, Eleventy, Tauri, DuckDB-WASM, Express, transformers.js |
| Go | Wails (Go + Svelte/React desktop apps) |
| Rust | Tauri (desktop apps) |
| Swift | iOS apps (Fly.io launcher, test apps) |
| GDScript | Godot game engine |
| SQL | PostgreSQL, DuckDB, SQLite |
Infrastructure & Platforms
| Category | Tools |
|---|---|
| Cloud | GCP Cloud Run, AWS (S3, Lambda, Route53), Fly.io, Modal, Supabase |
| Data | DuckDB, PostgreSQL, Neo4j, SQLite, Parquet, Overture Maps, Common Crawl, FRED API, DataBento, IBKR |
| AI APIs | OpenAI (GPT-4o, DALL-E), Anthropic (Claude), Google (Gemini), Groq (Llama), Ollama |
| LLM Tooling | MCP (Model Context Protocol), Claude Code, Pi Agent, Guidance, Outlines, LangGraph, PocketFlow, Burr |
| DevOps | Docker/OrbStack, git worktrees, Makefiles, uv, Playwright |
| Local Models | llama.cpp, Llama 3.2 1B, Gemma 2 2B, XTTS-v2, LFM2-350M (ONNX) |
Monthly Timeline
2024
Jul 2024 — Foundations
- OpenAI function calling & tool use patterns
- Overture Maps / DuckDB / Parquet exploration begins
- Guidance + llama.cpp constrained decoding
- Neo4j graph database exploration
- GraphRAG (Microsoft) exploration
- event_etl project started — NYC event data pipeline
- omnivore/ POC notebooks begin (web extraction, data collection, visualization)
Aug 2024 — Events & Speech
- Omnivore POCs continue (tool manager, state machine, newsletter ingestion, history)
- Mini-Omni speech model exploration
- Modal serverless testing
- Graph-based narrative simulation (sparse event graphs)
- Svelte + DaisyUI playground
- Burr state machine framework
- event_ui Svelte app started
- Gemma 2 2B local inference (Apple Silicon)
Sep 2024 — Data & Audio
- TTS deep-dive: Bark + XTTS models
- Overture Maps lat/lon correction pipeline
- DuckDB S3 parquet testing
- LightGBM exploration
- Omnivore: embeddings (Nomic), entity relationships, dossiers
Oct 2024 — LLMs from Scratch
- GPT Nano — Full nanoGPT implementation walkthrough
- Llama 3.2 E2E — Building Llama from scratch (Raschka)
- ALMOD experiments — Custom architecture modifications
- Spirit LM expressive model exploration
- Lightning LitGPT
- RL: Stable Baselines 3, PPO from scratch, Godot RL agents, DQN
- Browser-use automation library
- SNAC speech codec exploration
Nov 2024 — Models & Tableau
- tableau_model — PyTorch vision model training
- ALMOD continued (Tableau Nano variant)
- Algorithm design
- Spotify MCP server
- Volatility / reset experiments in play/
Dec 2024 — Time Series & Tricks
- Time series paper implementation (FRet/LoTRA)
- tricks — Plan→Act→Verify agent framework
- dlt tutorial
2025
Jan 2025 — Deep Learning Courses
- MIT Intro to Deep Learning course
- PyTorch Dataloader deep-dive
- JAX framework intro
- Modal testing
Feb 2025 — ML Ops & Experiments
- LLaDA — Large Language Diffusion model exploration
- Comet ML Ops, HuggingFace Datasets
- Lightning AI product analytics scraping
- mit_ml_course begins
- quick_experiments repo started
- inkanalytics.com updates
Mar 2025 — MCP & Agents
- MCP deep-dive (intro, OpenAI vs Anthropic comparison)
- agnt — Custom agent library
- taskwrapper_old — Early task management
- Browser-use testing
- PocketFlow, Pyro probabilistic programming
- ML play experiments
- tricks continued
Apr 2025 — Creative AI
- Google Sheets → Jupyter (Interhurr)
- llm_flow_game — LLM adventure with DALL-E scene generation
- miniblog (Astro) started
May 2025 — Probability & Images
- Probability fun, PyTorch learning
- Flux Schnell + Stable Diffusion image generation
Jun 2025 — Cloud & Probability
- gcp_llm_template — Cloud Run LLM deployment template
- PyMC probabilistic programming
- Puzzle experiments
Jul 2025 — Claude Code & Quant
- claude_code_project_template + cc_smol — Reusable agent scaffolds
- cc_jupydash — Tauri + Jupyter dashboard app
- cc_task_dash — Kanban + autonomous Claude agents
- cc_steering — Swarm intelligence + matrix cognition
- cc_moe — Spice Pirates game design
- quant repo — FRED, IBKR, DataBento, GARCH, pairs trading
- kalshi_exploration — Prediction market data
- cc_smol template
Aug 2025 — Mobile & Agents
- cc_ios_test — iOS Fly.io machine launcher with terminal
- cc_recursive — Desktop dev tools (Wails)
- Guidance docs, Ollama Python, kitten TTS
- pi-mono contributions begin
Sep 2025 — Games & Data
- cc_godot — Living Island RPG (Victorian + LLM)
- Common Crawl exploration (POC + Hiatusa)
- LangGraph, MCP refresher
- Matrix graph operations
- cc_ios_test continued
- cc_task_dash continued
- MIT ML course continued
Oct 2025 — Exchange & Skills
- exchange — Multi-service auction platform (FastAPI + SvelteKit)
- Math→Python (SymPy), SQL window functions
- Diffusion architecture diagrams
- skills (Anthropic) exploration begins
Nov 2025 — Gemini & Vision
- gemini_experiments — 14+ AI-generated HTML games
- gemini_tools — Narrative game engine with segmentation
- exchange_d — Exchange with Supabase
- sam3 — Segment Anything exploration
- docs — Mintlify documentation site
- Claude Code system prompts study
Dec 2025 — Fitness & Events v2
- circuit_duo — Partner circuit workout app
- exercise_db — Exercise database
- event_etl_revisited — AI-powered event pipeline v2
- pi-skills started
- Skills repo continued
2026
Jan 2026 — Business & Browser
- aao — Organization management platform
- cc_solar — Solar business orchestration
- tabreset — Chrome extension (local LLM tab summarizer)
- event_etl_revisited continued
- skill-building-experiments — Consulting/frontend/writing plugins
- event_ui updates
Feb 2026 — Pi Ecosystem & Solar
- pi-spot — VideoVault music video aggregator (agent-operated)
- pi-template / pi_solar_01 — Agent orchestrator templates
- agfact-monorepo — Agent factory
- miniblog continued
- pi-mono, pi-skills continued
- quick_experiments continued
*This document was auto-generated by scanning repo READMEs, git histories, and notebook contents.