Shu Lynn Liu
@shulynnliu
Followers
554
Following
432
Media
0
Statuses
64
CS PhD @BerkeleySky | Student Lead @NovaSkyAI Previously Undergrad @UWMadison | @utnslab | @mpi_sws_ Fan of @FCBayern #MiaSanMia
Berkeley, CA
Joined May 2020
🎯 AI discovers an algorithm that makes database transaction schedules 34% faster [ADRS Blog #4] We revisit a classic database problem: dealing with transactional contention. Starting with a state-of-the-art algorithm from our VLDB '24 paper, we use OpenEvolve to automatically
0
7
27
🧑🏫On-Policy Distillation is available as an example on SkyRL! The implementation required no library code changes, and we were able to reproduce AIME math reasoning experiments from the @thinkymachines blogpost. Check out our detailed guide to see how! https://t.co/TqDS649oFQ
Our latest post explores on-policy distillation, a training approach that unites the error-correcting relevance of RL with the reward density of SFT. When training it for math reasoning and as an internal chat assistant, we find that on-policy distillation can outperform other
5
5
23
Our group has multiple PhD student openings for Fall 2026, in Information Sciences and Computer Science. Welcome to apply! You may find details here:
vision.ischool.illinois.edu
The Computer Vision and Machine Learning Group conducts research at the intersection of computer vision and machine learning, with a focus on developing continual and data-efficient intelligent...
3
60
273
🚀 Introducing UCCL-EP: A portable, efficient Expert Parallelism framework that brings DeepEP-level GPU-driven communication with the same APIs to any cloud or hardware — AWS EFA, AMD GPUs, Broadcom NICs and beyond. Blog: https://t.co/d3oBVlWezZ Code: https://t.co/0UbCUYz9N9
1
6
7
🚀 AI for Systems is getting real! Our latest ADRS blog shows how LLMs can evolve new spot-instance scheduling algorithms, beating the NSDI’24 Best Paper algorithm in @SkyPilot_org! 💪Great work led by @andylizf and @tian_xia_. @andylizf is an amazing SkyLab undergrad intern
🎯 AI found algorithm beat * NSDI'24 Best Paper * 🤯 [ADRS Blog #2] We use AI to find new spot-instance scheduling algorithms. It beats the original paper algorithm by cutting cloud costs up to 48% (average 27%), while still meeting the job deadlines! ✍️ Read the blog:
0
3
13
☁️SkyRL now runs seamlessly with SkyPilot! Let @skypilot_org handle GPU provisioning and cluster setup, so you can focus on RL training with SkyRL. 🎯 Launch distributed RL jobs effortlessly ⚙️ Auto-provision GPUs across clouds 🤖 Train your LLM agents at scale Get started
0
10
23
🚀 SkyRL has day-zero support for OpenEnv!! This initial integration with OpenEnv highlights how easily new environments plug into SkyRL. Train your own LLM agents across containerized environments with simple, Gym-style APIs 🔥 👉 Check it out:
Excited to share OpenEnv: frontier-grade RL environments for the open-source community 🔥! https://t.co/KVeBMsxohL 🧩 Modular interfaces: a clean Gymnasium-style API (reset(), step(), state()) that plugs into any RL framework 🐳 Built for scale: run environments in containers
0
9
21
Kicking off a new ADRS blog series on how AI is transforming system research! The first blog dives into MoE load balancing, uncovering both algorithmic and engineering optimizations. Brilliant work done by @abmfy_. Check it out! 👇
🚀 We used AI to discover a new algorithm for LLM inference, achieving a 5.0x speedup in MoE load balancing over expert-written code. ✍️ Read the details in our blog post: https://t.co/sHVRqX6wDR 📄 Full paper: https://t.co/ex6AidUuwK 💻 Code: https://t.co/o2EVHmFMCl
0
2
8
🚀 End the GPU Cost Crisis Today!!! Headache with LLMs lock a whole GPU but leave capacity idle? Frustrated by your cluster's low utilization? We launch kvcached, the first library for elastic GPU sharing across LLMs. 🔗 https://t.co/3BC7B6s2EX 🧵👇 Why it matters:
9
53
198
🚀 AI is no longer just a "black box" for tuning systems. It's now a "white box" that rewrites core algorithms and outperforms human experts! We use AI-Driven Research for Systems (ADRS) frameworks to discover state-of-the-art solutions in under 12 hours for less than $20. This
adrs-ucb.notion.site
🗓️ Posted: October 17, 2025
0
7
25
Barbarians at The Gate: How AI is Upending Systems Research by @audreyccheng, @LynnLiu41887950, @melissapan, @istoica05, and the @ai4research_ucb team, https://t.co/b6vtMJN3Et This's first article of the The Next Horizon of System Intelligence blog series.
0
8
22
SkyRL v0.2.0 is here! With 22 contributors (including 11 new contributors!), this release holds many updates like strong MoE support with Megatron, LoRA support, standardized inference on the OpenAI API, new integrations, and many many more. The code: https://t.co/CWlKue6BU9
1
5
23
SkyRL just crossed 1000 Github stars! Thank you to all the wonderful contributors and users building this project together 🥳 Check it out: https://t.co/CWlKue79JH
0
6
38
Reading between the lines, this whole trend is an incredible opportunity for the PL research community. The next generation of systems research will be one level of abstraction higher than it is today. This is what PL research is all about (especially OOPSLA / PLDI crowd).
Barbarians at the Gate is a very interesting new paper, with some exciting results showing the potential for AI in systems research. But I think the authors aren't quite asking the hardest problem about where this takes systems as a field. I wrote a new blog post about it.
2
5
38
That's a fair point: systems research is often about abstractions and trade-offs, not just better metrics! We see AI-driven optimization in systems performance as a first step: it explores broader design spaces, sometimes discovering ideas humans might miss. The exciting part
@ai4research_ucb Do you really see systems research as an exercise in hill climbing? I don’t think AI is upending systems research, just incremental systems work. If anything the focus should be (has always been) on real innovation in the problem formulation, the tradeoffs, and abstractions —
0
2
10
I forgot about this tweet but read this top tier paper and get ultra agi pilled.
Few questions those who are following AlphaEvolve and FunSearch * is anyone reproducing it? * very relevant to diverse data generation in verifiable domains? * one step away from a new paradigm beyond current thinking: “solve this problem under x constraint”? 1. Makes use of
5
19
252
We've been getting some great results with AI for systems optimization at Berkeley! This is a way of using AI that I think is under-explored: use LLMs as a prior to search for a solution given a verifiable metric. It could probably be applied to more industry problems.
🚀 Excited to release our new paper: “Barbarians at the Gate: How AI is Upending Systems Research” We show how AI-Driven Research for Systems (ADRS) can rediscover or outperform human-designed algorithms across cloud scheduling, MoE expert load balancing, LLM-SQL optimization,
2
19
92
3->5, 4->6, 9→11, 7-> ? LLMs solve this via In-Context Learning (ICL); but how is ICL represented and transmitted in LLMs? We build new tools identifying “extractor” and “aggregator” subspaces for ICL, and use them to understand ICL addition tasks like above. Come to
6
36
211