I lead 0-1 design and deployment for Poly, originally building SOTA foundation models for design assets.
My best work revolves around model behavior, spanning ♠ human data, ♥ user studies,
♦ prompt design, and ♣ interaction design.
Real datasets and links included. Reach out to see full case studies or design work!
LLM Fine-Tuning
,
2025
Poly is currently building an intelligent cloud file system. We're redesigning the traditional file storage experience to give you smarter, more native ways to search, browse and organize your files. The objective was to fine-tune an AI agent that understands all your media, so users can find anything in natural language, summarize a folder of documents, automatically organize, tag & rename files, and more. TLDR Key Initiatives: LLM tone dissection study, prompt library creation, actionable user studies
LLM Fine-Tuning
,
2025
Poly is currently building an intelligent cloud file system. We're redesigning the traditional file storage experience to give you smarter, more native ways to search, browse and organize your files. The objective was to fine-tune an AI agent that understands all your media, so users can find anything in natural language, summarize a folder of documents, automatically organize, tag & rename files, and more. TLDR Key Initiatives: LLM tone dissection study, prompt library creation, actionable user studies
SVG Model
,
2023
We built a next-generation SVG model that enabled designers to generate vector assets from text, sketches, reference images, or even extrapolate full asset sets from a single style example. TLDR Key Initiatives: Large scale data labeling projects, novel interaction patterns, actionable user studies
SVG Model
,
2023
We built a next-generation SVG model that enabled designers to generate vector assets from text, sketches, reference images, or even extrapolate full asset sets from a single style example. TLDR Key Initiatives: Large scale data labeling projects, novel interaction patterns, actionable user studies
3D Textures Model
,
2022-23
Poly started when we built the first SOTA model for 3D texture generation — allowing designers to create production-ready AI-generated textures. These were PBR materials in 8K resolution and 32-bit EXR maps, from just a photo or text prompt, and customizable in a full editor. With over 200K monthly active users and 100K+ daily generations, we also ran a large scale creator program producing human-generated training data as inputs to our reward model pipeline. TLDR Key Initiatives: Hand curated dataset, human data generation
3D Textures Model
,
2022-23
Poly started when we built the first SOTA model for 3D texture generation — allowing designers to create production-ready AI-generated textures. These were PBR materials in 8K resolution and 32-bit EXR maps, from just a photo or text prompt, and customizable in a full editor. With over 200K monthly active users and 100K+ daily generations, we also ran a large scale creator program producing human-generated training data as inputs to our reward model pipeline. TLDR Key Initiatives: Hand curated dataset, human data generation
Voice Biomarkers
,
2021-22
Kintsugi is developing novel voice biomarker infrastructure to score clinical depression and anxiety from 20-seconds of free form speech. My role involved establishing data acquisition strategies contributing to the world's largest voice dataset annotated for mental health foundation models, leading the vetting and outreach process to 100+ clinical trial recruitment companies and cross-promotional partners. Project details under NDA
Voice Biomarkers
,
2021-22
Kintsugi is developing novel voice biomarker infrastructure to score clinical depression and anxiety from 20-seconds of free form speech. My role involved establishing data acquisition strategies contributing to the world's largest voice dataset annotated for mental health foundation models, leading the vetting and outreach process to 100+ clinical trial recruitment companies and cross-promotional partners. Project details under NDA