ML engineer & interpretability researcher in NYC — real-time SAE-steered diffusion, feature visualization for DINOv2, subspace geometry in diffusion transformers.https://hammamiomar.com/https://hammamiomar.com/blog/dino-dreamer/https://hammamiomar.com/blog/dino-dreamer/A work-in-progress feature-visualization instrument for navigating frozen representation models — DINOv2, an archetypal SAE, and SteerViT. An honest checkpoint: what I built, what I believe, and what I still cannot claim.Tue, 23 Jun 2026 00:00:00 GMThttps://hammamiomar.com/blog/subspacelens/https://hammamiomar.com/blog/subspacelens/I applied MCR² coding rate metrics to a frozen PixArt-α and found progressive subspace separation, dense activations, and text-independent structure — evidence that a subspace lens may complement SAE-based interpretability.Mon, 13 Apr 2026 00:00:00 GMThttps://hammamiomar.com/blog/fifty-fps-realtime-sdxl/https://hammamiomar.com/blog/fifty-fps-realtime-sdxl/The engineering behind Latent Dance — SAE-steered, audio-reactive diffusion at 50 FPS on a single RTX 5090Fri, 06 Mar 2026 00:00:00 GMThttps://hammamiomar.com/blog/letting-computers-dance/https://hammamiomar.com/blog/letting-computers-dance/Music visualization, generative models, and building a tool to see what's inside.Fri, 06 Mar 2026 00:00:00 GMThttps://hammamiomar.com/blog/drums-sae/https://hammamiomar.com/blog/drums-sae/A quick experiment in interpretable audio featuresMon, 12 Jan 2026 00:00:00 GMThttps://hammamiomar.com/blog/learning-llm/https://hammamiomar.com/blog/learning-llm/A reflection on llm usage while learning at The Recurse CenterSat, 31 May 2025 00:00:00 GMThttps://hammamiomar.com/blog/return-statement/https://hammamiomar.com/blog/return-statement/My return statement for Spring 1 2025 Recurse Center batchSat, 31 May 2025 00:00:00 GMT