Skip to content
View Amine-LG's full-sized avatar
🎯
Focusing
🎯
Focusing
8 followers · 3 following
  • Red Hat Certified in OpenShift · K8s Contributor

Achievements

Achievement: Pull Sharkx2Achievement: Starstruckx2Achievement: YOLOAchievement: Pair ExtraordinaireAchievement: Quickdraw

Achievements

Achievement: Pull Sharkx2Achievement: Starstruckx2Achievement: YOLOAchievement: Pair ExtraordinaireAchievement: Quickdraw

Block or report Amine-LG

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Amine-LG/README.md

Amine Largou, independent software developer in Morocco: I keep asking software why. A git-style graph of my work: the main line, trading systems in MQL5 for clients, runs from 2017 to today. AI and LLMs branch off in 2024, Kubernetes and OpenShift platforms in 2025, and decision models in 2026. Every line is still active.

Independent developer in Morocco, about nine years in. I build trading systems that run unattended in live markets and, increasingly, Python automation, Kubernetes and OpenShift platforms, and AI tools. I like software best when it can show its work.

aminelg.com  ·  Red Hat certified  ·  MQL5

Selected work

Lineage

Who can access what in OpenShift, and why, including access that quietly comes back. Live demo →

Why can alice read secrets in payments-prod? The live path: user alice, group engineers, RoleBinding secret-readers, Role read-secrets, which allows get and list on secrets. Access that can come back: a ClusterRoleBinding still names a deleted ServiceAccount, ci/deployer. If someone recreates that name, the old grant is live again without anyone approving it twice. Lineage flags it before then.

IntentSQL

Ask a SQLite database a question in plain English. The SQL is written by code, never by a model.

One recorded IntentSQL answer. Question: which five districts have the highest per-pupil expenditure? Code proposes readings, Jev chooses between them, code writes the SQL, Jev checks the result. Jev chose: show the district name, 87%; show per-pupil expenditure, 55%; read five as LIMIT 5, 100%. Code wrote a query joining districts and expenditures, ordered by per-pupil expenditure, limit 5. Jev's check was unsure, so the answer is flagged uncertain and the other reading is offered too.

Jev is TypeSafe's decision model. Instead of generating text, it picks from options it's given and returns a probability for each. That's why IntentSQL can show every choice, and flag the answers it isn't sure about.

More

  • Obsidian Text Transformer: a local LLM turns plain notes into structured Obsidian Markdown, on your own machine.
  • Decision Lab: give several decision models the same input, then change the evidence and see what flips.
  • Jev Creature Forge: a sentence becomes a typed creature genome, drawn as SVG by deterministic code.
  • Aim Duel: a real-time 1v1 browser game on my own Kubernetes cluster, deployed with GitOps. Play it.
  • MultiVision Dashboard and PatternVision Pro: MetaTrader 5 tools for watching many markets at once and scanning for candle patterns.

Upstream

# Kubernetes docs: Pod failure policy
-  …taking at least 2 minutes.
+  …this example would take at least
+  9 minutes before the Job fails.

I ran the example and it took 9 minutes 6 seconds. More fixes merged into the Kubernetes docs, plus contributions to pytest and the OpenShift docs.

Toolbox

python:  [automation, fastapi, flask]
linux:   [podman, docker, bash]
k8s:     [kubernetes, openshift, helm]
gitops:  [argo-cd, github-actions]
ai:      [local-llms, ollama, agents]
web:     [js, node, react, svelte]
trading: [mql5, metatrader-5]

certified: [EX280, EX180]  # Red Hat

Have a system that needs building, or explaining? Tell me about it.

Pinned Loading

  1. Obsidian-Text-Transformer-LocalAI Obsidian-Text-Transformer-LocalAI Public

    Transform your .txt and .md notes into visually appealing and well-structured Markdown for Obsidian, powered by local LLMs using Ollama.

    Python 129 10

  2. lineage lineage Public

    Who has access to what in OpenShift, and why — including the grants that survive deletion.

    Python 3 1

  3. Discord-Ollama-ChatBot Discord-Ollama-ChatBot Public

    Simple Discord Ollama ChatBot

    Python 8

  4. IntentSQL IntentSQL Public

    Natural-language SQLite interface and experiment with Jev / System One decision models: inspectable query plans, deterministic SQL, and guarded writes.

    Python 4 1

  5. jev-creature-forge jev-creature-forge Public

    Typed semantic creature compilation with Jev: natural language → inspectable genome → deterministic procedural SVG.

    Python 1

  6. aim-duel aim-duel Public

    Real-time multiplayer workload on Kubernetes with GitOps, WebSockets, TLS ingress, readiness probes, and a Redis/HPA scaling roadmap.

    JavaScript