Insights, updates, and stories from the future of sneaker design
January 15, 2025
Reflections on scaling from 50 to 1,500 users. Technical deep dive on inference optimization, model serving architecture, and maintaining sub-60s latency at scale. Lessons from distributed GPU cluster management and cost optimization strategies.
December 18, 2024
Technical breakdown of our geometry generation system. Signed distance functions vs occupancy networks, differentiable rendering for 2D-to-3D training, custom loss functions for manufacturing constraints, and topology optimization post-processing. Includes ablation studies and performance benchmarks.
November 8, 2024
How we enable rapid fine-tuning on client design libraries with minimal data. Meta-learning protocols, contrastive learning for style embeddings, and ensemble methods for consistent brand output. Technical analysis of our 150M parameter style encoder architecture.
September 22, 2024
Production deployment results from major footwear brand. Quantitative analysis of workflow improvements, quality metrics comparison vs traditional CAD, and integration challenges with existing PLM systems.
July 12, 2024
Creators can now sell their designs directly to brands through our new marketplace. Building the infrastructure for design licensing, version control, and attribution tracking.
June 3, 2024
Adapting latent diffusion for footwear design. Domain-specific conditioning mechanisms, custom loss formulations for geometric feasibility, and techniques for maintaining material compatibility during sampling. Includes training curves and evaluation metrics.
May 5, 2024
How we achieve production-ready mesh generation without 3D training data. Differentiable rendering enables end-to-end learning from 2D images. Technical deep dive on render loss formulations, normal consistency terms, and silhouette matching for geometric priors.
March 18, 2024
Why we chose to bootstrap ShoeLabs instead of raising venture capital. Our approach to sustainable growth, maintaining control, and building a business that serves users first.
February 1, 2024
Opening beta access to 50 designers. Initial model architecture, training methodology, and validation results. Early performance metrics and user feedback from first production deployments.
December 10, 2023
Why existing generative models fall short for industrial design. Technical analysis of manufacturing constraint satisfaction, material realism requirements, and 3D geometric accuracy needs that define the footwear design problem space.
November 15, 2023
Why we're building AI for footwear design. Analysis of the design workflow gap, technical feasibility assessment, and our hypothesis for bringing generative models to industrial 3D design.