[0;32m➜[0m [1;36mincognito@quack[0m:[1;34m~[0m$ ./init --stealth
[0;90m[ OK ][0m mounting /dev/curiosity
[0;90m[ OK ][0m loading keypair ........................ [0;32mdone[0m
[0;90m[ OK ][0m shredding metadata ..................... [0;32mdone[0m
[0;90m[ WARN ][0m caffeine buffer at [0;33m12%[0m
[0;90m[ OK ][0m spawning 27 repositories ............... [0;32mdone[0m
[1;32m __
(o > [0m[0;37mQUACK. Identity redacted. Code is not.[0m[1;32m
\\ )
--"-[0m
[0;32m➜[0m [1;36mincognito@quack[0m:[1;34m~[0m$ [5m█[0m
🔍 Expand: how I actually use each one →
| Layer | Weapons | What I ship with it |
|---|---|---|
| Frontend | JavaScript · HTML5 · CSS3 · Tailwind | Interfaces that stay out of your way |
| Backend | Node.js · Express · Socket.io | Realtime rooms, ephemeral sessions, file relays |
| Security | Web Crypto API · AES · hashing · zero-log design | E2E chat, encrypted transfer, secret sharing |
| AI/ML | Python · PyTorch · Hugging Face · scikit-learn | Small models, RAG pipelines, on-device inference that doesn't phone home |
| Data | MongoDB · Firebase · localStorage-first | Store the minimum. Forget the rest. |
| Ops | Git · GitHub Actions · Netlify · Vercel | Push → build → live in under a minute |
🧠 Expand: the mental model I build with →
%%{init: {'theme':'base','themeVariables':{'primaryColor':'#0d1117','primaryTextColor':'#00ff9c','primaryBorderColor':'#00ff9c','lineColor':'#2dd4bf','fontFamily':'JetBrains Mono'}}}%%
flowchart LR
A([💡 idea at 2AM]) --> B{does it leak data?}
B -- yes --> C[redesign it]
C --> B
B -- no --> D[[prototype in vanilla JS]]
D --> E{does it feel fast?}
E -- no --> F[delete half the code]
F --> E
E -- yes --> G[[encrypt everything]]
G --> H([🚀 deploy to netlify])
H --> I([🦆 quack])
Splitting time between shipping full-stack apps and getting models to behave — same privacy-first rules apply.
{
"focus": ["local-first LLM tooling", "RAG pipelines", "fine-tuning small models", "on-device inference"],
"stack": ["Python", "PyTorch", "Hugging Face", "scikit-learn", "Jupyter", "Pandas/NumPy"],
"philosophy": "If the model needs your data to leave the device, it's not private — it's rented."
}No public AI/ML repos pinned yet — check back, or open an issue if you want to compare notes.
🗂️ Expand the full case file →
| # | Project | Classification | Payload |
|---|---|---|---|
| 01 | SecureChat | realtime · e2e |
Anonymous rooms with unique join codes, end-to-end encrypted, nothing persisted. Node + Socket.io. |
| 02 | Extensions | browser · tooling |
ElabXCS — a browser extension that makes Elab actually bearable. |
| 03 | File-sharing | transfer |
Direct, no-account file relay between users. Node backend, dead-simple front end. |
| 04 | Finance-Manager | tooling · local-first |
Track money without handing it to a third party. |
| 05 | Random-Projects | sandbox |
The lab. Experiments, prototypes, half-ideas that sometimes graduate. |
| 06 | Research-Paper | archive |
Open-access collection of papers worth actually reading. |
Live deployments
- 🔐
encryptera.netlify.app— encryption playground - 🧪
securecode.netlify.app— secure code utilities
🔓 Terminal challenge — decode this →
$ echo "UVVBQ0tfRkxBR3t5b3VfZm91bmRfdGhlX2hpZGRlbl9sYXllcn0=" | base64 -dRun it. If you get the flag, open an issue titled QUACK and I'll follow you back. Seriously.
🦆 Ask the duck a question →
Q: "Why isn't my code working?"
__
(o > A: "You know why. Read line 47 again."
\\ )
--"-
Q: "But it worked yesterday—"
__
(o > A: "Yesterday you had a semicolon."
\\ )
--"-
Q: "Should I just use an LLM for this?"
__
(o > A: "Only if it forgets you afterward."
\\ )
--"-
Rubber-duck debugging is free. Consider me on call.
📊 My commit hours, honestly →
🌞 Morning ▓░░░░░░░░░░░░░░░░░░░░░░░ 4%
🌆 Daytime ▓▓▓▓▓░░░░░░░░░░░░░░░░░░░ 19%
🌃 Evening ▓▓▓▓▓▓▓▓░░░░░░░░░░░░░░░░ 31%
🌙 Night ▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░░░░░ 46% ← the good stuff
"Anonymity isn't hiding. It's refusing to be the product."
⭐ If any of this was useful, star a repo. It's the cheapest dopamine on the internet.
{ "alias": "IncognitoQuack", "role": "Full-stack developer · privacy & security tinkerer · AI/ML tinkerer", "stack": ["JavaScript", "Node.js", "Python", "Socket.io", "HTML/CSS"], "ml_stack": ["PyTorch", "Hugging Face", "scikit-learn", "Pandas/NumPy", "Jupyter"], "obsessions": ["end-to-end encryption", "realtime systems", "clean UX", "on-device inference"], "shipping": ["Encryptera", "SecureCode", "SecureChat", "Finance Manager"], "philosophy": "The best data you can leak is the data you never stored — models included.", "status": "compiling something you're not supposed to know about" }