Stack

What I use, what I want to learn next, and what I've dropped.

Stack
Name	Purpose	Thoughts	Status
Docker	Containers	Can't ship anything without it, so no complaints here.	Active
LiteLLM	Model routing	Handles my routing, and sometimes I use it to run Chinese models with Claude Code. Very useful.	Active
Tailwind CSS	Styling	Styling without leaving the markup. It's on almost everything I build and I haven't looked back.	Active
Supermemory	AI memory	Way easier to integrate than Mem0, and the features keep shipping, which I like.	Active
Pydantic	Data validation	Shows up in almost everything I write in Python. Hard to remember life before it.	Active
Figma	Design	Still where I sketch out UI before building anything.	Active
LangGraph	Agent orchestration	Has almost everything you need to take an AI product to production. Best of the bunch for me.	Active
Redis	Caching	Somehow always ends up in my stack.	Active
Hugging Face	Model hub	Where I go to try new lightweight models and deep learning projects. The datasets are the best part.	Active
Next.js	Web framework	My default for anything on the web.	Active
Claude Code	Coding agent	Best for architecting and the more complicated stuff. Most of my heavier work goes through it.	Active
Neon	Serverless database	Spins up fast and is perfect for side projects.	Active
Playwright	Browser automation	Reliable for testing and scraping.	Active
scikit-learn	Classical ML	Still helps on deep learning projects, just not as much as when ML was all I did.	Active
Bun	JS runtime	Very fast, though I don't feel much pull toward it since Anthropic acquired it.	Active
Framer Motion	Animation	Makes adding motion easy without much effort.	Active
Celery	Task queue	Takes the heavy jobs off the API so nothing hangs.	Active
Python	Main language	Great for most of my AI work. Slow, but not as slow as JS, so I manage.	Active
Langfuse	LLM observability	One suite for testing, observation, and versioning. Works really well.	Active
Mermaid	Diagrams	Fastest way to get an architecture diagram into a doc.	Active
PostgreSQL	Database	Does almost everything I need, so why look elsewhere?	Active
OpenRouter	Model gateway	Great for production apps. Swapping models is easy and I get analytics too.	Active
Zod	Validation	Keeps inputs and outputs honest on the TypeScript side.	Active
Kubernetes	Orchestration	Reach for it when a project outgrows a single box.	Active
PyTorch	Deep learning	Best framework out there and easier than TensorFlow. Most of my deep learning projects run on it.	Active
Better Auth	Authentication	Easy to set up and does the job without any drama.	Active
Ghostty	Terminal	Fast and clean, and I'm in it every day.	Active
shadcn/ui	UI components	Clean UI without fighting a design system.	Active
Codex	Coding agent	Very budget friendly, especially with ChatGPT Luna.	Active
FastAPI	Backend framework	My go-to for backends. Quick to build with and fits the AI stack well.	Active
Elasticsearch	Search engine	Powerful, scales well, and great for full-text, log, and vector search. The learning curve is steep and it eats memory though.	Active
TypeScript	Main language	Great for quick prototyping. Don't really use it for much else.	Active
PydanticAI	Agent framework	Very powerful, just not LangGraph powerful. Good for prototyping fast.	Active
Supabase	Backend platform	Quickest way to get a backend running without much setup.	Active
Superset	Agent multitasking	I love multitasking, so I mostly want to see how well it supports that.	Planned
MLflow	Experiment tracking	Curious how it compares to the other debug, evaluate, and monitor tools.	Planned
OpenTelemetry	Observability	Been meaning to learn it properly instead of half understanding it.	Planned
Qdrant	Vector database	Mostly just trying to find an alternative to pgvector because I'm bored.	Planned
Rust	Systems language	Very interested since it's so memory efficient, but still thinking about that steep learning curve.	Planned
Go	Backend language	Just curious.	Planned
Kafka	Event streaming	Still learning it and want to build something real with it.	Planned
DuckDB	Local analytics	Want to try it for quick analysis without spinning up a whole database.	Planned
Terraform	Infrastructure code	Learning it so I can stop clicking around cloud dashboards.	Planned
Prometheus	Monitoring	Want to learn it alongside Grafana.	Planned
Grafana	Dashboards	Want to learn it so I can actually see what my systems are doing.	Planned
Helm	Kubernetes packaging	Next up once I'm comfortable with Kubernetes.	Planned
Tree-sitter	Code parsing	Want to understand how code actually gets parsed.	Planned
Cursor	Code editor	Used it for a while, but it slowly fell out of my rotation.	Inactive
OpenCode	Coding agent	Was nice and still is, but I settled on Codex and Claude Code since I'm more into their ecosystems.	Inactive
Gemini CLI	Coding agent	Plainly sucks.	Inactive
Zed	Code editor	Nice and fast, but extension support is limited for now. Might switch back later.	Inactive
Ollama	Local models	Very useful, my laptop just can't handle local models.	Inactive
Qwen	Open-source models	Used it for projects and work and it's wonderful. I just don't need it right now.	Inactive
TensorFlow	Deep learning	Great for very advanced deep learning, but I moved over to PyTorch.	Inactive
LangChain	LLM framework	Shifted toward agentic products instead of chain-style ones.	Inactive
RAGAS	RAG evaluation	Used it for a while, then moved to Langfuse.	Inactive
R	Data analysis	Used it heavily for data work until I realised pandas and Python's plotting libraries could do the same.	Inactive
React	UI library	Still around through Next.js, I just don't touch it on its own anymore.	Inactive
Three.js	3D graphics	Fun to play with, but no reason to use it right now.	Inactive
Node.js	JS runtime	Python and Bun cover most of what I need now.	Inactive
SSE	Response streaming	Used it for streaming responses, but I don't need it much at the moment.	Inactive
Neo4j	Graph database	Tried it, but haven't had a project that needed a graph lately.	Inactive
SearXNG	Metasearch	Used it for search in a few projects, nothing needs it right now.	Inactive
Crawlee	Web scraping	Used it for scraping, but nothing calls for it at the moment.	Inactive
Linear	Issue tracking	Never really became part of my workflow.	Inactive