People really underestimate how much work and practice are required to get good at something. A truckload of work over a long period of time is the price of competence.
Ikeja Electric was hacked through a profile photo upload form.
The threat actor exploited an Unrestricted File Upload vulnerability on their Smart Warehousing Inventory Management System, uploaded a webshell, and had remote access within minutes.
From there, he moved through their network, finding passwords stored in plain text, using them to access internal systems, and eventually cracking the domain admin password.
Gaining full control in four days.
He then exploited an unpatched VMware vCenter server—software from 2018, never updated—and according to him, deployed ransomware across 50+ hosts, taking down metering software across their systems.
I’ve published a full analysis of what I’m now calling cyber-terrorism against Nigerian critical infrastructure, along with a practical advisory for affected individuals, organisations, and regulators.
https://t.co/0rzvhVcLNf
I trained a 12M parameter LLM on my own ML framework using a Rust backend and CUDA kernels for flash attention, AdamW, and more.
Wrote the full transformer architecture, and BPE tokenizer from scratch.
The framework features:
- Custom CUDA kernels (Flash Attention, fused LayerNorm, fused GELU) for 3x increased throughput
- Automatic WebGPU fallback for non-NVIDIA devices
- TypeScript API with Rust compute backend
- One npm install to get started, prebuilt binaries for every platform
Try out the model for yourself: https://t.co/TB2itlmCVT
Built with @_reesechong. Check out the repos and blog if you want to learn more.
Shoutout to @modal for the compute credits allowing me to train on 2 A100 GPUs without going broke
cc @sundeep@GavinSherry