Mohamed Yasser

Solution Architect, Sharjah City Municipality

[Mohamed Yasser @yasser_rmd]

@paulabartabajo_ Thank you for expressing exactly what I was thinking. You’ve put it into words beautifully.

[Mohamed Yasser @yasser_rmd]

Tested @Kimi_Moonshot 3D with photos of our Sharjah City Municipality Head Office. It took 3 attempts to get the curved wings right.

𝗡𝗼𝘁 𝟭𝟬𝟬% 𝗮𝗰𝗰𝘂𝗿𝗮𝘁𝗲, but promising for rapid 3D prototypes in municipal construction reviews.

@KimiDevs #kimi

[Elvar @elvar_official]

https://t.co/XgMNZnz7E3

[Mohamed Yasser @yasser_rmd]

@datachad @ollama 0.5 for now, demo only. Plan is a lower threshold plus escalation on low confidence, tuned on labeled cases.

[Mohamed Yasser @yasser_rmd]

Part 2 of my System One experiment with @ollama.

Most AI safety checks are quick judgment calls: block or allow, escalate or not.

A local model that returns a choice with a probability fits that well. Next: a framework around it.

[Mohamed Yasser @yasser_rmd]

Tried @ollama System One + @bespokelabsai Nimble in a live warehouse simulation.
The model is making real-time operational decisions like reroute, wait and return-to-charge in ~200–350 ms locally.
This is where small decision models get interesting.
#Ollama #SystemOne #LocalAI

[ollama @ollama]

Ollama now supports Jev-like decision models all locally.

Use decision models like Nimble for tasks like ticket triaging, model routing, and content moderation.

ollama pull nimble

Here’s Nimble playing Ollama racer through the new local /v1/systemone API by making decisions in real-time. 🏎️

[Mohamed Yasser @yasser_rmd]

@UnslothAI @danielhanchen @Nandakishorm1, is the one who is build this model

[Mohamed Yasser @yasser_rmd]

@agenticgirl Nice one. I built something in a similar space around 9 months back called Barq-GraphDB, focused on graph + vector storage for AI agents, with agent decision tracking and audit trails built in.

https://t.co/OXRjWppe7d

[Mohamed Yasser @yasser_rmd]

ỷ̷͖̺͔̩̬̭̫͓̾̾̾͗̋͂̓̅͝ö̷̙̘̲̗͕̰͓͖͕̺̼̪́u̶͍͍͉̤͎̇̾̍̇͗͛̄̉̚ ̴̛̤̯͑̒̀͝á̵̘̳̘̱͎̙̟͉̿̏̂̍̈́̾͗ṟ̴̂͋̑̉̅͐͊͆͐̏̀͠e̷̢̨̟͖̯̥̦̬͍̿̀̂ͅ ̸̡̨̝̻̰͎̥̰̆̈̈́̋̈̾͒́̕ͅs̵̤̩̭̤̯̟͈͎̙̒̀̈́̔̏̎͌c̶̬͊̀a̵͔̖͌̎̀͊̍̍͑̚͜r̶̨̧̲̦̘̻̲̥͖̙̳͓͌̈́̂̔̈̚ȩ̸̢̯̬͕̥͍̺͙̻̰̀̀̽̈́̅̎ḏ̵̞̫͋̇͂͋͐̾́́͠ͅ ̵̥͙̇̑̽̆̏

[Mohamed Yasser @yasser_rmd]

For prompt check out here https://t.co/FEQB0GHv0g

[Mohamed Yasser @yasser_rmd]

Stress-tested Gemini Omni Flash across 6 languages with ultra-fast cinematic transformations, live environments, lip sync, and dynamic audio.

Very impressive overall. Minor word repetition hiccups under extreme pacing, but surprisingly smooth.

#GeminiOmni #GoogleAI #AIVideo

[Mohamed Yasser @yasser_rmd]

Codex + @Blender test: text → coastal villa + animation.

4-bed, double-height, infinity pool, sunset lighting.

Fast concept to 3D.

#AI #Blender @OpenAI

[Mohamed Yasser @yasser_rmd]

Built an open source tool that verifies your AI token bills. Providers tell you the count. Nobody checks. siqlah does.

https://t.co/B8uOV1le3Q

https://t.co/ZR4IlYT582

#AI #OpenSource #LLM

[adam @adamdotnew]

Opus 4.7 appears to be SOTA at agentic CAD design

[Pau Labarta Bajo @paulabartabajo_]

Wanna learn with me how to build a wildfire prevention system using a Small Vision Language Model?

Next week I am hosting a 60-minute technical deep dive.

Where? 🧵(1/3)

[Pau Labarta Bajo @paulabartabajo_]

Advice for AI engineers 💡

You don't need a trillion-parameter model to change the world.

A fine-tuned VLM like LFM2-VL + satellite imagery is enough.

10 real-world problems you can solve today 🧵↓

[Mohamed Yasser @yasser_rmd]

repo: https://t.co/A87LH40yon

[Mohamed Yasser @yasser_rmd]

I built this after realizing SQL generation is the easy part.

Understanding the schema and executing queries is what actually matters.

Schema-Forge
A schema-aware SQL coding agent.

Works across databases.

Link in reply

#AI #SQL #Agents #OpenSource

[Pau Labarta Bajo @paulabartabajo_]

What if you could catch illegal fishing from space?
Or track plastic drifting across the Pacific?
Or map urban heat islands before the next heatwave kills?

April 13th: build it.

Using @liquidai LFM2/2.5-VL models and @DPhiSpace satellite images.

$20K+ in prizes.

Register today ↓
https://t.co/JXwrCnNZex

[Mohamed Yasser @yasser_rmd]

repo: https://t.co/sMqwtFORlv

[Mohamed Yasser @yasser_rmd]

Built AiMesh, a Rust-based message queue for AI agents.
Cost-aware routing, budget control, semantic dedup, scatter-gather workflows.
Target: 5M+ msg/sec, sub-ms latency.
AI infra needs more than simple API calls.
#AI #Rust #LLM #DistributedSystems #AIInfrastructure