Applied Scientist

Block MB; City Of London; Hybrid; GBP 60,000 - 80,000; Full time

Job summary

A small graphics-focused startup in London is seeking an Applied Research Scientist to drive the development of models for generating high-quality 3D assets from real-world images. You will propose and evaluate innovative techniques in physically-based rendering and generative AI. Ideal candidates should possess a PhD or equivalent experience along with a strong background in PyTorch and generative methods. The role offers competitive salary, equity, and a flexible hybrid work setup.

Benefits

Competitive salary

Equity

Flexible hybrid setup

Qualifications

• PhD (or equivalent experience) in ML, computer vision, graphics or a closely related field.
• Strong PyTorch skills and experience training models on large‑scale image or 3D data.
• Track record in generative AI and material estimation.

Responsibilities

• Propose and evaluate models for PBR material prediction and 3D asset creation.
• Build on modern vision transformers and diffusion architectures.
• Work closely with engineering teams to implement models in production tools.

Skills

Strong PyTorch skills

Experience in generative AI

Comfortable taking ideas from sketch to prototype

Experience training models on large-scale image data

Education

PhD in ML, computer vision, graphics or related field

Job description

Applied Research Scientist – 3D, PBR & Generative Models

London preferred | Full-time | Visa sponsorship available

The mission

Help turn messy, real‑world photos into production‑ready, PBR‑accurate 3D assets using generative models. This is a small, graphics‑focused startup working with high‑end film, games and immersive studios, building tools that quietly sit behind some very pretty pixels.

The role

You’ll be the researcher driving the models that predict material properties and generate geometry from multi‑view image data. Your work won’t live in a PDF – it will ship into real artist workflows, fast, and help automate the jump from photos to textured, physically‑based assets.

What you’ll do

• Propose, implement and evaluate state‑of‑the‑art models for PBR material prediction and 3D asset creation.
• Build on top of modern vision transformers and diffusion‑style architectures to extract maps like roughness, normals, specular and diffuse from images.
• Design models that respect multi‑view consistency and can work from dense shots through to sparse or text‑conditioned inputs.
• Run experiments, define baselines and track progress against product‑level quality bars, not just academic metrics.
• Work closely with engineering teams to get your models into production tools and influence how a unique 3D / PBR dataset is collected and used.

What you bring

• PhD (or equivalent experience) in ML, computer vision, graphics or a closely related field.
• Strong PyTorch skills and experience training models on large‑scale image or 3D data.
• Track record in generative AI, material estimation, 3D‑aware learning or similar high‑impact research areas.
• Comfortable taking ideas from sketch → prototype → evaluated model, quickly.
• Bonus: photogrammetry, material / UV workflows, PBR and shading models, self‑supervised methods (e.g. CLIP‑style), 3D reconstruction, inverse rendering or texture synthesis.

Why you might care

• Work at the crossroads of 3D graphics, physically‑based rendering and generative AI, using rare, real‑world PBR capture data rather than synthetic assets.
• Collaborate directly with artists, engineers and studios who will rely on your models to remove manual texturing from their workflows.
• Competitive salary, equity, flexible hybrid setup in London.