Location:Remote
È la persona giusta per questa opportunità? Si assicuri di leggere la descrizione completa qui sotto.
Project Type:Freelance / Contract
The Opportunity: Join a Rocket Ship
AIVBisn't just a startup; we are a high-velocity scaleup redefining how AI solutions reach the market.
Hyper-Growth: 300% YoY growth, jumping from zero to £1.4M in our first full year.
Proven Pedigree: Founded by the pioneers behind Deep Learning Italia, the gold standard for AI education in Italy.
Global Ambition: Backed by £4.5M in seed funding, we are preparing for a major US-led Series A in Q4 2026 to dominate international markets.
We are building the "Operating System" for AI ventures. We don't just build products; we build companies.
About AIVB
Founded in 2024 by Marco Tognoni and Matteo Testi, AIVB is an AI Venture Builder designed to scale innovation at speed. Our team of 40 is anchored by an elite engineering core, primarily AI PhDs, who transform complex research into vertical‐specific powerhouses.
We operate through a Centralized Powerhouse that develops strategic blueprints for specialized spinoffs in high-impact sectors such as Fashion, Real Estate, and CyberDefense. We are "International by Design", English is our language, and global scale is our target.
Project Overview
We are looking for a hardcore Machine Learning Engineer / Data Scientist to architect and deploy the backend pipeline for a cutting‐edge 2D/3D fashion generation platform.
We are building a generative AI platform that synthesizes sketches, 3D garment models, and technical drawings by ingesting brand archives, mood boards, and historical sales data. You will not be designing the clothes; you will be building the custom, on‐premise ML infrastructure that enables our platform to do so.
Crucially, this is an entirely on-premise, code‐first pipeline. You must be capable of deploying, fine‐tuning, and managing open‐source models (via Python, PyTorch, and Diffusers) locally on GPU clusters, without relying on commercial cloud APIs.
Core Engineering Responsibilities
Architecting Text-to-2D Pipelines: Design and deploy robust diffusion model pipelines using Python and the Hugging Face Diffusers library, moving beyond basic UI usage.
Developing Conversational Editing Workflows: Programmatically implement workflows for iterative image manipulation (e.g., integrating ControlNet, IP-Adapter, and specialized inpainting models into our backend architecture).
2D-to-3D Model Deployment: Implement, optimize, and serve state‐of‐the‐art AI models (Trellis3D, TripoSR, Hunyuan3D) to convert 2D generated assets into 3D meshes programmatically.
Local Infrastructure & VRAM Optimization: Manage model weights, optimize inference times, and handle VRAM constraints using techniques like xFormers, FlashAttention, TensorRT, and FP16/BF16 quantization.
Dataset Preparation & Fine-Tuning: Build automated pipelines to fine‐tune custom models (LoRAs, Dreambooth) on structured fashion datasets, ensuring the outputs mathematically respect physical realities like fabric drape and structural integrity.
Required Technical Expertise
Educational Background: Master's or Ph.D. in Computer Science, Artificial Intelligence, Machine Learning, or a closely related technical field.
Professional ML Experience: At least 3 years of industry experience working explicitly as a Data Scientist, Machine Learning Engineer, or AI Researcher.
Programming & Frameworks: Strong proficiency in Python, PyTorch, CUDA basics, and the Hugging Face ecosystem.
Diffusion Models: Deep, under‐the‐hood knowledge of the 2D generation ecosystem (Stable Diffusion 1.5, SDXL, SD3, Flux) and custom fine‐tuning methodologies.
2D to 3D Generation: Hands‐on programmatic experience with modern 3D generation models (Trellis3D, TripoSR, Hunyuan3D).
On‐Premise Management: Proven ability to run complex pipelines locally. While familiarity with UIs like ComfyUI is great for prototyping, you must be able to translate these into scalable, headless backend code.
Domain Context: A foundational understanding of fashion terminology and garment construction to effectively evaluate and guide model outputs.
Nice to Have (Bonus Skills)
(Please note: Proficiency in the following software is highly valued, but without the underlying Python/PyTorch machine learning skills outlined above, your application will not be considered.)
Experience with Clo3D or Marvelous Designer: for garment simulation.
Proficiency in Blender: for mesh cleanup, retopology, and scripting AI-generated 3D assets via Python APIs. xrdztoy
Are you ready to shape the future of AI in fashion? Apply now!