GLM 5.3 uncensored setup with browser and full agent tools

We need an experienced AI engineer to get GLM 5.3 running on a rented GPU server in an uncensored / unrestricted build (or the closest available GLM release if 5.3 is not out yet), and to give the model a full tool stack so it can operate on its own. What the setup needs: - Browser control (Playwright or Chrome DevTools Protocol): open pages, click, type, read content - Shell / code execution - File read and write - Web search and page fetching - Any tool-calling framework you prefer (we are flexible on the stack) Deliverables: 1. A working deployment on a server we can access, running for a 4 hour session 2. The model responding without refusals (uncensored / abliterated build, or your recommended approach) 3. Tool use wired in and demonstrated live: the model opens a browser, navigates and completes a simple task end to end 4. A short handover: how to restart it, how to plug in new tools, and the running costs Requirements: hands-on experience with vLLM, llama.cpp or SGLang, GPU hosting (RunPod, Vast, Lambda, AWS), quantization, and agent tool frameworks. Please share one or two similar deployments you have done. Fixed price: $100, one-time. Intermediate level. We will leave a five-star review on successful delivery.