Mohamed Bal

Founder & CTO of DEVUP AI — Algeria’s AI infrastructure platform for 200+ models, unified APIs, GPU/CPU cloud compute, and local DZD billing.

Algiers, Algeria Joined on Jul 1, 2026 https://devupdz.com

Work

Founder & CTO at DEVUP AI

Mohamed Bal

Sep 19

Build a Streaming AI Chatbot with Next.js, Vercel AI SDK, and DEVUP AI

#nextjs #ai #typescript #tutorial

11 min read

Mohamed Bal

Sep 19

Connect DEVUP AI to VS Code—No Extension Required

#vscode #ai #programming #tutorial

9 min read

Mohamed Bal

Sep 13

Claude Code in Algeria: A Practical Setup with Local DZD Billing

#ai #programming #tutorial #claudecode

8 min read

Mohamed Bal

Jul 15

Your SAE Passed the Cosine Similarity Bar. That Doesn't Mean It

#machinelearning #deeplearning #ai #llm

6 min read

Mohamed Bal

Jul 10

A Perfect 1.000 Cosine Match That Never Fires: The Hidden Trap in SAE Evaluation

#ai #python #opensource #machinelearning

26 min read

Mohamed Bal

Jul 8

I Ran a Causal Test on Sparse Autoencoders — 77% of 'Recovered' Features Turned Out to Be Causally Inert

#ai #machinelearning #python #llm

26 min read

Mohamed Bal

Jul 4

Quantum-Inspired, Not Quantum: The Physics of Tensor Networks Behind Production LLM Compression

#ai #machinelearning #python #performance

23 min read

Mohamed Bal

Jul 4

MiniMax-M3 on DEVUP AI — A Technical Deep Dive into MiniMax Sparse Attention, Benchmarks, and Production Integration

#ai #llm #api #buildinpublic

7 min read

Mohamed Bal

Jul 1

Why Algeria Needs Its Own AI Infrastructure — and Why I'm Building It

#ai #webdev #opensource #machinelearning

3 min read

DEV Community

Mohamed Bal

Badges

Writing Debut

Skills/Languages

Currently learning

Currently hacking on

Available for

Build a Streaming AI Chatbot with Next.js, Vercel AI SDK, and DEVUP AI

Connect DEVUP AI to VS Code—No Extension Required

Claude Code in Algeria: A Practical Setup with Local DZD Billing

Your SAE Passed the Cosine Similarity Bar. That Doesn't Mean It

A Perfect 1.000 Cosine Match That Never Fires: The Hidden Trap in SAE Evaluation

I Ran a Causal Test on Sparse Autoencoders — 77% of 'Recovered' Features Turned Out to Be Causally Inert

Quantum-Inspired, Not Quantum: The Physics of Tensor Networks Behind Production LLM Compression

MiniMax-M3 on DEVUP AI — A Technical Deep Dive into MiniMax Sparse Attention, Benchmarks, and Production Integration

Why Algeria Needs Its Own AI Infrastructure — and Why I'm Building It