Hi there, I'm Yanis Miraoui
- 🪐 I work on diffusion based LLM models at Inception since 2024. Contact me, we're hiring!
- 🇺🇸 I graduated from a MS in Statistics at Stanford University in 2025. 🌉
- ⚡ I am mostly interested in Problem Solving and Machine Learning.
- 🎲 In my free time, I love playing and watching sports. ⚽️
- BenchLeader — Compare AI models across published benchmarks, speed, latency, and cost in one leaderboard.
- PennyOCR — Turn PDFs and images into text and Markdown with an OCR API built for developers and AI agents.
- Tensorlight — Explore AI GPUs, memory, and inference through interactive 3D models and visual explanations.
- PaperKit — Merge, split, crop, sign, and convert PDFs with tools that process documents locally in your browser.
- Idiomesh — Generate native Python and TypeScript SDKs from OpenAPI. Try the live compiler; hosted automation is in private beta.
- GitShine — Schedule GitHub commits, customize their messages and cadence, and discover projects through the community.
- Unmark — Clean hidden Unicode characters from text and rewrite AI-generated prose with adjustable strength.
- OPT Pulse — Explore community-reported OPT and STEM OPT timelines, compare similar cases, and estimate processing windows.
- Hourra! — Bring wearable data together to explore sleep, recovery, activity, and longer-term health trends.
- Voiceflint — Build voice agents for phone and web with flexible model choices and clear call costs and latency.
- Everscene — Turn Apple TV into an ambient display, with an iPhone companion for creating and sharing AI scenes.
- Zapline — A native IPTV player for your own playlists on iPhone, iPad, Mac, and Apple TV, with a program guide and catch-up playback.
- Mercury: Ultra-Fast Language Models Based on Diffusion — Co-authored the report introducing Mercury, a family of diffusion language models that generates tokens in parallel for fast text generation. (2025)
- SpaRG: Sparsely Reconstructed Graphs for Generalizable fMRI Analysis — Co-first author. Combines sparsification and self-supervision to identify informative brain connections that generalize across fMRI acquisition sites. (MICCAI 2024) · Code
- NLP Research project: Analyzing sports commentary to automatically recognize events and extract insights ⚽⛹️
- ↪️ With its demo 💻
- ↪️ ArXiv article
