You are being re-directed to my new personal website in 3 seconds.
- Future Distributed Machine Learning Systems, 2018 - Current
Distributed machine Learning (ML) systems are constantly challenged by new models, such as large NLP models, Graph Neural Networks (GNNs), Reinforcement Learning (RL) and Federated Learning (FL).
Their designs also need to evolve with emerging hardware platforms, such as mobile robots and trusted enclaves.
In this project, we explore for new distributed ML systems [VLDB 2019, HotCloud 2020] that can effectively support new models and exploit new hardware platforms.
- Self-driving Real-time Data Analytics Systems, 2017 - Current
Big data companies, such as Microsoft, Google and Facebook, need to process data ingested from global cloud servers and edge devices.
The data ingestion workload is highly dynamic and contains massive unstructured/semi-structured data.
In this project, I am designing distributed real-time analytic systems [VLDB 2018] that can automatically adapt their configurations using ML techniques.
- Future Data Centre Networks for Big Data Computation, 2012 - 2017
Big data applications, such as distributed machine learning and data analytics, often suffer from scaling bottlenecks in cloud data centres. In this project, we pioneered the idea of exploiting application-specific in-network computing for scaling Big data applications.
To fully achieve the promise of this idea, we designed new network architectures [CoNEXT 2014, HotCloud 2015, HotCloud 2016]
and network programming techniques [USENIX ATC 2017, USENIX ATC 2016].
I am leading the development of the following open-source software:
I am fortunate to supervise/co-supervise the following talented students:
Andrei-Octavian Brabete, MEng, Imperial College London, 2019. (Next destination: G-Research)
Ioan Budea, MEng, Imperial College London, 2019. (Next destination: Facebook)
For a full list of my papers, please check my Google scholar page.
Spotnik: Designing Distributed Machine Learning for Transient Cloud Resources
Marcel Wagenlander, Luo Mai, Guo Li, Peter Pietzuch
The 12th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud), 2020
Adaptive Distributed Training of Deep Learning Models
Luo Mai, Guo Li, Andrei-Octavian Brabete, Alexandros Koliousis, Peter Pietzuch
The AI System Workshop in Symposium on Operating Systems Principles (SOSP), 2019
[Invited oral presentation]
Taming Hyper-parameters in Deep Learning Systems
Luo Mai, Alexandros Koliousis, Guo Li, Andrei-Octavian Brabete, Peter Pietzuch
ACM SIGOPS Operating Systems Review, July, 2019
[Invited vision paper]
CrossBow: Scaling Deep Learning with Small Batch Sizes on Multi-GPU Servers
Alexandros Koliousis, Pijika Watcharapichat, Matthias Weidlich, Luo Mai, Paolo Costa, Peter Pietzuch
Proceedings of Very Large Data Base (VLDB), 2019
Chi: A Scalable and Programmable Control Plane for Distributed Stream Processing Systems
Luo Mai, Kai Zeng, Rahul Potharaju, Le Xu, Shivaram Venkataraman, Paolo Costa, Terry Kim, Saravanan Muthukrishnan, Vamsi Kuppa, Sudheer Dhulipalla, Sriram Rao
Proceedings of Very Large Data Base (VLDB), 2018
TensorLayer: A Versatile Library for Efficient Deep Learning Development
Hao Dong, Akara Supratak, Luo Mai, Fangde Liu, Axel Oehmichen, Simiao Yu, Yike Guo
ACM Multimedia 2017
[Best Open-source Software Award]
[Invited talks: Google London DevFest 2018, London AI Symposium 2018]
Emu: Rapid Prototyping of Networking Services
Nik Sultana, Salvator Galea, David Greaves, Marcin Wojcik, and Jonny Shipton, Richard Clegg, Luo Mai, Pietro Bressana and Robert Soule, Richard Mortier, Paolo Costa, Peter Pietzuch, Jon Crowcroft, Andrew W Moore, Noa Zilberman
The USENIX Annual Technical Conference (ATC), 2017
Towards a Network Marketplace in a Cloud
Da Yu, Luo Mai, Somaya Arianfar, Rodrigo Fonseca, Orran Krieger, David Oran
The 8th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud), 2016
FLICK: Developing and Running Application-specific Network Services
Abdul Alim, Richard G. Clegg, Luo Mai, Lukas Rupprecht, Eric Seckler, Paolo Costa, Peter Pietzuch, Alexander L. Wolf, Nik Sultana, Jon Crowcroft, Anil Madhavapeddy, Andrew Moore, Richard Mortier, Luis Oviedo, Masoud Koleni, Derek McAuley, Matteo Migliavacca
The USENIX Annual Technical Conference (ATC), 2016
Optimizing Network Performance in Distributed Machine Learning
Luo Mai, Chuntao Hong, Paolo Costa
The 7th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud), 2015
NetAgg: Using Middleboxes for Application-specific On-path Aggregation in Data Centres
Luo Mai, Lukas Rupprecht, Abdul Alim, Paolo Costa, Matteo Migliavacca, Peter Pietzuch, Alexander L. Wolf.
The 10th ACM International Conference on emerging Networking EXperiments and Technologies (CoNEXT), 2014
[Best paper finalist]
- Exploiting Time-malleability in Cloud-based Batch Processing Systems
Luo Mai, Evangelia Kalyvianaki, Paolo Costa
The 7th ACM SIGOPS Workshop on Large-Scale Distributed Systems and Middleware (LADIS'13) co-located with SOSP, 2013
[Media highlight: The Register]
Load Balanced Rendezvous Data Collection in Wireless Sensor Networks
Luo Mai, Longfei Shangguan, Chao Lang, Junzhao Du, Zhenjiang Li, and Mo Li.
The 8th IEEE International Conference on Mobile Ad-hoc and Sensor Systems (MASS), 2011
[Best paper finalist]
- Visiting Researcher, Azure Big Data Group, Microsoft, USA, 2017 - 2018
- Research Intern, Azure Big Data Group, Microsoft, USA, 2016
- Research Intern, Microsoft Research Redmond, USA, 2015
- Research Fellow, Microsoft Research Asia, China, 2014
- MRes in Advanced Computing (Distinction Honour), Imperial College London, UK, 2011 - 2012
- BSc in Computer Science (China National Scholarship Honour), Xidian University, China, 2007 - 2011.
- Alibaba Innovative Research Award, 2020
- Microsoft Azure Research Award, 2018
- Best Open-source Software Award, ACM Multimedia Conference, 2017
- Google Fellowship in Cloud Computing, 2012 - 2016
- Best Paper Finalist, ACM CoNEXT Conference, 2014
- Best Paper Finalist, IEEE MASS Conference, 2011