Full-Stack Developer ยท AI/LLM Engineer ยท Security Researcher
Developer with a focus on LLM infrastructure, model deployment, and web development.
Interested in the intersection of open-source AI tooling, backend systems, and application security.
AI & LLM
- Model inference & serving โ vLLM, LM Studio, Ollama
- Quantization โ GPTQ, AWQ, FP8 dynamic (see HuggingFace โ)
- LLM APIs โ OpenAI, Anthropic Claude, Mistral
- Prompt engineering, RAG pipelines
Infrastructure & DevOps
- AWS EC2 โ deployment & inference of large models
- Docker โ containerization of ML workloads
- Linux server administration
Web Development
- Backend โ Laravel, Node.js
- Frontend โ Next.js, Vue.js, SvelteKit
- Systems โ Rust (Axum)
prism-php/prism โ Unified LLM interface for Laravel (2.3k โญ)
| PR | Type | Description |
|---|---|---|
| #748 | Feature | Added keep_alive option for Ollama provider |
| #687 | Bug fix | Fixed response parsing when output contains ```json blocks |
| #403 | Feature | Added multimodal support to PrismServer API endpoints |
CVE-2024-47186 โ XSS in Filament (Laravel full-stack framework)
Unvalidated color values passed to ColorColumn / ColorEntry components allowed XSS injection via Laravel's @style Blade directive. Affected versions v3.0.0โv3.2.114. Patched in v3.2.115 following responsible disclosure. CVSS 3.1: 6.1 (Medium)
| Model | Task | Downloads |
|---|---|---|
| Voxtral-Small-24B-2507-FP8-dynamic | Automatic Speech Recognition (24B) | 26 |
HuggingFace ยท GitHub ยท Linkedin