AI / Python Developer for Multi-Agent 3D Scanning & Analysis Workflow (Contract to Full-Time)

Project Overview: We are an innovative startup based in Hong Kong building a fully automated, AI-driven pipeline to manufacture custom 3D-printed insoles. We are looking for a highly skilled technical partner to build out this platform in phases. This initial contract is for Phase 1 (Proof of Concept & Initial Workflow). If Phase 1 is successful, this role will transition into a long-term, full-time partnership to build out the remaining phases (CAD generation and 3D printing integration). Phase 1 Scope of Work: You will be responsible for building a multi-agent AI workflow that takes user-uploaded media, converts it into a 3D model, and extracts precise anatomical measurements. The specific workflow you will build: User Interface: A simple chat interface where a user can upload a video or a batch of images of their foot. Agent 1 (Vision & Orchestration): Receives the files and validates that there are enough angles/frames for a successful 3D scan. Sends the validated media to a 3D reconstruction API (e.g., Luma AI, Polycam, or similar). Retrieves the processed 3D mesh file (.obj or .ply). Saves the file to a database and passes the file path/context to Agent 2. Agent 2 (Data Analyst & Geometry): Uses Python 3D libraries (such as trimesh or open3d) to programmatically extract physical measurements from the 3D foot mesh (e.g., arch height, foot width, foot length). Generates 2D renders/snapshots of the 3D mesh. Passes these 2D renders to a Vision LLM (like GPT-4o or Claude 3.5 Sonnet) to visually analyze the foot and identify biomechanical traits like pronation or supination. (Note: We understand pressure points are difficult to map purely from a mesh without a pressure mat, so we are looking for your expertise on how to best estimate weight distribution based on arch type and foot shape). Required Skills & Experience: Python Programming: Strong backend Python skills. AI Agent Frameworks: Experience building multi-agent workflows using frameworks like CrewAI, LangGraph, AutoGen, or Dify. 3D Geometry / Computer Vision: Experience working with 3D meshes using libraries like trimesh, open3d, or pyvista. API Integrations: Experience integrating external APIs (OpenAI, Anthropic, Luma AI, etc.). Basic UI/UX: Ability to spin up a quick front-end (e.g., Streamlit, Gradio, or a basic React/Next.js app) to demonstrate the chat and upload interface. Future Phases (For the right long-term partner): Once Phase 1 is complete, future work will involve: Agent 3: Generating automated, professional podiatry reports based on the data. Agent 4 (CAD/CAM): Using parametric modeling (e.g., OpenSCAD, KittyCAD API) to automatically generate a 3D-printable .stl file of the custom insole based on Agent 2's measurements. To Apply, Please Answer the Following: Briefly describe how you would approach building Agent 1 (validating the images and interacting with a 3D API). Have you worked with 3D files (.obj, .stl, .ply) in Python before? If so, which libraries did you use? Which multi-agent framework (CrewAI, LangGraph, etc.) would you recommend for this specific workflow and why? Please provide a rough time and cost estimate for Phase 1 as described above