Ryan Arman, PhD — Machine Learning & Generative AI Engineer
Machine learning engineer and technical lead (15+ years) who builds and ships production generative AI — LLM fine-tuning and distillation, AI agents, and retrieval-augmented generation (RAG) — alongside large-scale ranking and award-winning medical AI. Experience at Google, Meta, Amazon, and AI startups (two as founder/CEO). PhD (University of British Columbia) and Stanford postdoc; IPCAI 2012 Best Paper Award.
Kirkland / Seattle, WA · LinkedIn · Google Scholar · Contact via the form on arman.bio
Experience
- Machine Learning Engineer, Oumi (enterprise LLM fine-tuning platform), 2025–present — LLM fine-tuning and post-training; on-policy distillation compressing near-frontier quality into small deployable models; contributed to an in-platform LLM agent; rigorous LLM-as-judge evaluation; sub-1B models that beat frontier LLMs at a fraction of inference cost.
- Machine Learning Engineer, Meta (Ads Ranking), 2024–2025 — models and continuous-training pipelines powering $50B+ annual ad revenue; A/B experimentation under privacy constraints.
- Founder / CEO, Yora AI, 2023–2024 — on-device, privacy-preserving RAG product to search a user's own files (Mac / Windows).
- Machine Learning Engineer / Technical Lead, Google (Health AI Research), 2019–2023 — deployed deep-learning co-pilot for radiologists (lung-cancer detection from chest CT), embedded with clinicians and a hospital partner; NLP and multimodal models.
- Software Development Engineer, Amazon Music, 2017–2019 — deep-learning search ranker in a Tier-1 service (10,000+ requests/second).
Selected project
FilmEase (filmease.net) — a solo-built, live generative-AI movie-recommendation product: retrieval over movie embeddings feeds an LLM that reranks to the user's taste and explains each pick (agentic RAG), across web, iOS, and Android.
Selected publications
- Evaluating the Google CT Foundation Model for central pulmonary embolism detection on CT pulmonary angiograms — European Journal of Radiology (2026). link
- Assistive AI in lung cancer screening: a retrospective multinational study in the United States and Japan — Radiology: Artificial Intelligence (2024). doi
- Artificial intelligence as a second reader for screening mammography — Radiology Advances (2024), Best of Radiology Advances 2024. doi
- Feasibility of image registration for ultrasound-guided prostate radiotherapy via a convolutional neural network — Technology in Cancer Research & Treatment (2019). doi
- Single-camera closed-form real-time needle tracking for ultrasound-guided needle insertion — Ultrasound in Medicine & Biology (2015). doi
- A closed-form differential formulation for ultrasound spatial calibration — IPCAI (2012), Best Paper Award.
Education
- Postdoctoral Research, Stanford University School of Medicine (2016)
- PhD, Electrical & Computer Engineering, University of British Columbia (2014)
- MSc, Electrical Engineering, K. N. Toosi University of Technology (2009); BSc, University of Tehran (2007)