Ana Klimovic
Assistant Professor
Builds cloud computing and storage systems, including efficient data pipelines and resource management for large-scale machine learning training and serving.
Ana Klimovic's lab at ETH Zurich works on efficient training & inference and distributed, cloud & edge systems. 34 papers since 2023, 17 at top venues, mostly SOSP and VLDB. 5 PhD students and 5 master's students. 36 alumni, most often to Apple and Google DeepMind.
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 50%
- PhD students
- 5
- MS students
- 5
- alumni listed
- 36
Papers by yearfrom the lab homepage, 2023–2026
- 5 papers2026 · T1 3 · collecting
- 13 papers2025 · T1 8
- 6 papers2024 · T1 3
- 10 papers2023 · T1 3
T1 venuesOtherVenues: SOSP, VLDB, EuroSys, ICML, MLSys
Research focuspaper titles that use each topic's words
- Efficient training & inference9 papers
- Distributed, cloud & edge systems7 papers
- builds cloud0 papers
- pipelines0 papers
- large-scale machine0 papers
Paper list
From the lab homepage.
20265 papers
- ACL (T1)Apertus: Democratizing Open and Compliant LLMs for Global Language Environments
- CIDREnd-to-End Declarative Data Analytics: Co-designing Engines, Interfaces, and Cloud Infrastructure
- SIGMOD (T1)Mixtera: A Data Plane for Foundation Model Training
- OSDI (T1)OpenTela: Unifying Decentralized Computing Resources for Heterogeneous LLM Serving (Operational Systems)
- Proceedings of the ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOSTaming the Long-tail: Efficient Reasoning RL Training with Adaptive Drafter
202513 papers
- ATC[ATC] Resource Multiplexing in Tuning and Serving Large Language Models
- CIDRAdaptive Data Transformations for QaaS
- EuroSys (T1)DeltaZip: Efficient Serving of Multiple Full-Model-Tuned LLMs
- ICML (T1)Demystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs
- SIGMOD (T1)Modyn: Data-Centric Machine Learning Pipeline Orchestration
- MLSys (T1)On Distributed Larger-Than-Memory Subset Selection With Pairwise Submodular Functions
- ASPLOS (T1)PCcheck: Persistent Concurrent Checkpointing for ML
- SoCCRethinking Web Cache Design for the AI Era
- SOSP (T1)Sailor: Automating Distributed Training over Dynamic, Heterogeneous, and Geo-distributed Clusters
- MLSys (T1)ThunderServe: High-performance and Cost-efficient LLM Serving in Cloud Environments
- SoCCTowards a Lightweight Sidecar-based Service Mesh for Serverless
- SoCCUnderstanding GPU Resource Interference One Level Deeper
20246 papers
- Proceedings of the Workshop on Hot Topics in System Infrastrcutrue, co-located with SOSPDecluttering the data mess in LLM training
- ICML (T1)DéjàVu: KV-cache Streaming for Fast, Fault-tolerant Generative LLM Serving
- SOSP (T1)Dirigent: Lightweight Serverless Orchestration
- CIDROff-the-shelf Data Analytics on Serverless
- EuroSys (T1)Orion: Interference-aware, Fine-grained GPU Sharing for ML Applications
- ATCPecan: Cost-Efficient ML Data Preprocessing with Automatic Transformation Ordering and Hybrid Placement
202310 papers
- VLDB (T1)Analyzing Vectorized Hash Tables Across CPU Architectures
- Proceedings of the Workshop on Resource Disaggregation and Serverless (WORDS) at SOSPEnabling In-Vitro Serverless Systems Research
- VLDB WorkshopEphemeral Per-query Engines for Serverless Analytics
- SoCCFunction as a Function
- VLDB (T1)NVM: Is it Not Very Meaningful for Databases?
- VLDB WorkshopRethinking Serverless Computing: from the Programming Model to the Platform Design
- VLDB (T1)SHiFT: An Efficient, Flexible Search Engine for Transfer Learning
- SoCCtf.data service: A Case for Disaggregating ML Input Data Processing
- EuroMLSysTowards a Platform and Benchmark Suite for Model Training on Dynamic Datasets
- Proceedings of the Workshop on Resource Disaggregation and Serverless (WORDS) at SOSPUnderstanding the Neglected Cost of Serverless Cluster Management
Where alumni went36 listed, 100% to industry
Industry 4Academia 0
Most often to
- Apple
- Google DeepMind
- NVIDIA
Apply to ETH ZurichRolling
Not announced yet
Computer Science Doctorate (D-INFK)
- Round
- Rolling (no deadline)
- Opens
- —
- GRE
- See program page
- English
- Proficiency in written and spoken English
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- USUCSD
Hao Zhang
- Efficient training & inference
- CNSJTU
Qinya Li
- Distributed, cloud & edge systems
- SGNTU
Chen Wang
- Distributed, cloud & edge systems
- CNSJTU
Chao Li
- Accelerators, chips & architecture
- USUT Austin
Neeraja Yadwadkar
- Efficient training & inference
- CNPeking
Xuanzhe Liu
- Distributed, cloud & edge systems
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.