3D-to-AI Automation: Build a Cloud Product Render Generator

We are seeking an experienced 3D/AI Automation Engineer or Machine Learning Visualizer to build an automated image-generation pipeline and deliver our first batch of product catalog assets. We are working with a luxury architectural lighting collection (Brokis Arcade). We have all the original structural files (3DS, DWG, SKP) and 2D technical spec sheets. However, the manufacturer lacks exhaustive product photography for all of their modular variations. This is a two-phase project. We want to start by generating 360 flat 2D image variations for a single collection (9 light configurations \times 40 material/finish combinations). You will build the automation pipeline, deliver a simple no-code cloud interface for us, and output the first 360 images. The Material Training (WordPress Access Provided) The core challenge is capturing the "luxury physics" of premium hand-blown Bohemian glass (translucency, light refraction, saturation on curved edges), alongside wood and high-end metal finishes. To make training easy, we have a WordPress backend with an extensive library of Brokis lifestyle photos. Crucially, these images have already been clearly labeled in the file names (e.g., images containing "amber glass", "smoke glass", etc., are already tagged). You will be given access to pull these diverse reference images to train your dataset. Phase 1: Pilot Project Scope & System Build 1 Automated Structure Export: Use a script (Blender Python / 3ds Max script) to open our 9 structural files and automatically export perfectly identical, unwarped front/three-quarter black-and-white wireframe snapshots to use as our ControlNet base. 2 Material Library Setup: Use the pre-labeled WordPress reference images to train specialized machine learning components (custom LoRAs / IP-Adapter weighting) so the AI perfectly isolates and renders the distinct glass refractions and metal finishes onto separate zones of the lighting fixtures. 3 The Web Interface: You will host this pipeline on a clean, secure cloud platform (e.g., ComfyUI via Runflow, Comfy Cloud, RunPod, or a simple custom web UI). The interface must allow us to easily select a light, select a color variant via drop-down, and feature a one-click "Regenerate" button to randomize light reflections on the fly without writing text feedback. 4 Deliverables: The completed web tool setup + the first pilot batch of 360 high-res, photorealistic 2D renders organized neatly in a Google Drive folder. Phase 2: Scale Retainer (270+ Products) Once the system and material library are locked in and verified during Phase 1, we will transition into a long-term agreement. You will be retained to update the engine with new 3D models so we can run the pipeline across the rest of our 270+ product catalog, paying a highly optimized flat execution fee. Application Requirements (To be considered, you must answer): 1 What specific AI framework (e.g., ComfyUI, Stable Diffusion XL/Flux, specialized LoRAs, IP-Adapters) do you propose using to handle premium, translucent glass textures without altering the 3D geometry? 2 Confirm that you can build or host a clean, cloud-based web interface where we can execute/regenerate images without needing advanced hardware locally. 3 Please provide a fixed-price quote for Phase 1 (The build + the first 360 images).