Contrarian
I implement AI across every department to cut busywork, tighten execution, and help businesses scale with less overhead.
Current Focus
AI Implementation
AI Across Every Department
End-to-end AI implementation — from vision models to voice agents to full SaaS platforms.
End-to-end virtual staging pipeline built on open-source vision models. Feed it an empty room photo and it segments the space with SAM3, generates depth maps with Depth Anything, then composites photorealistic furniture using fine-tuned LoRA diffusion models — swapping layouts, lighting, and decor styles in minutes. The whole pipeline runs locally, no API costs per image.
How the pipeline works
Raw Photo
MLS listing image
Segmentation
SAM isolates regions
Depth Map
Depth Anything v2
Inpainting
Nano Banana + FLUX.2
Staged
Photorealistic result
Things I've Built
Conversational real estate search. Instead of filtering by bedrooms and price, describe your life — "I'm a chef with 2 kids, need a big garden and a short commute downtown" — and the engine matches you to listings using vector embeddings over property descriptions, neighborhood data, and spatial features.
Best Parlay
3-Leg · NFL/NBA
+285
Market
12%
Model
34%
Edge
+22pp
8yr NFL/NBA data
Chrome extension that automatically bookmarks everything you browse, auto-deletes entries after 7 days, and sends a weekly email digest summarising what you've been reading — so nothing slips through and your bookmarks never pile up. Shelved after Chrome changed their extension publishing algorithms.
The Stack
Tools I actually use to ship AI systems.
Agent Architecture
Frameworks
Image & Video Generation
Computer Vision
Voice AI
Languages
Outreach Platforms
IDEs / Coding Tools
Recent Thinking
Ready to scale?
Ready for 2026. Let's build something that converts.