Jinwon Lee - Meta | LinkedIn (original) (raw)
About
Experienced AI/ML Tech Lead and Manager with a proven track record of delivering…
Activity
Experience & Education
Meta
View Jinwon’s full experience
See their title, tenure and more.
Publications
Patents
Honors & Awards
#1 in CVPR'19 Visual Wake Words Challenge
-
#3 in NeurIPS'19 Micronet Challenge
-
Best Demo Award in IEEE SECON'12 Conference
-
Best Paper Award in IEEE MDM'06 Conference
-
Outstanding Paper Award in MLSys 2025
-
Qualcomm Annual Technology Conference (QTech'15)
-
Qualcomm Innovation Fellowship (QIF'19) Mentor Recognition
-
More activity by Jinwon
I am pleased to share that I have started a new position at LinkedIn, where I will be concentrating on AI Training Infrastructure. This marks an…
Liked by Jinwon Lee
How does the system behave when running SOTA agentic AI? Agentic AI is growing in complexity: multi-model, multi-tool, and complex autonomous…
Liked by Jinwon Lee
🎉 Excited to share that our paper, "Pushing the Limits of Block Rotations in Post-Training Quantization", has been accepted to ICML 2026! This one…
Liked by Jinwon Lee
𝐕𝐋𝐌³: 𝐕𝐢𝐬𝐢𝐨𝐧 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥𝐬 𝐚𝐫𝐞 𝐧𝐚𝐭𝐢𝐯𝐞 3𝐃 𝐥𝐞𝐚𝐫𝐧𝐞𝐫𝐬 3D understanding still runs on expert vision models:…
Liked by Jinwon Lee
Reasoning in Memory (RiM) enables latent reasoning without the need to "think out loud." By reasoning directly within a dedicated latent…
Liked by Jinwon Lee
People treat 5D parallelism (DP/CP/TP/PP/EP) as 5 independent knobs. Once sequence lengths vary, it isn't. Variable-length training turns hybrid…
Liked by Jinwon Lee
Would you like to join the research effort on JEPA and World Models easily? After a full year of hard work, we’re excited to finally release…
Liked by Jinwon Lee
After wonderful years at MIT as a postdoc, I'll be joining the Department of Computer Science and Engineering at Penn State University as a…
Liked by Jinwon Lee
𝐎𝐧-𝐝𝐞𝐯𝐢𝐜𝐞 𝐌𝐢𝐱𝐭𝐮𝐫𝐞-𝐨𝐟-𝐄𝐱𝐩𝐞𝐫𝐭𝐬 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥 We just released MobileMoE, the first sub-billion-active-parameter…
Liked by Jinwon Lee
Just read Google's latest paper on their Self-Evolving Recommendation System deployed at YouTube, and the technical architecture is genuinely…
Liked by Jinwon Lee
96% of edge LLM inference time isn't compute or DRAM. It's flash memory transfer. A new University of Florida paper (𝐂𝐀𝐓𝐒) shows why EAGLE…
Liked by Jinwon Lee
🧲 Why SFT, RL, and On-Policy Distillation? Nice blog post that compares SFT, RL, and on-policy distillation as different ways of reshaping a…
Liked by Jinwon Lee
Instead of watching an hour movie, watch this. In 14 minutes, an Anthropic engineer who wrote Building Effective Agents will teach you more about…
Liked by Jinwon Lee
Google just released an entire open-source library of production-grade AI agent examples — and it’s seriously impressive. This isn’t a collection of…
Liked by Jinwon Lee
Other similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
Others named Jinwon Lee in United States
20 others named Jinwon Lee in United States are on LinkedIn