Dan Alistarh
Alistarh earned an EPFL PhD with Rachid Guerraoui and later conducted research at MIT, Microsoft and ETH before joining ISTA. His work ranges from algorithmic bounds to practical machine-learning systems. Alistarh coauthored SparseGPT, SparseProp and other 2023 studies on sparse models and pruning. These investigate ways to reduce neural-model computation while preserving useful behavior. His work includes 2024 research on quantized models, fine-tuning and compact optimization, followed by 2025 work including Marlin and low-precision training.
- Now at
- Institute of Science and Technology Austria and Red Hat
- Works
- 287
- Citations
- 4,053
- h-index
- 30
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
As of October 2026
Theory and systems research
Alistarh earned an EPFL PhD with Rachid Guerraoui and later conducted research at MIT, Microsoft and ETH before joining ISTA. His work ranges from algorithmic bounds to practical machine-learning systems.
-
2023
Sparse neural computation
Alistarh coauthored SparseGPT, SparseProp and other 2023 studies on sparse models and pruning. These investigate ways to reduce neural-model computation while preserving useful behavior.
-
2024–2025
Compressed language models
His work includes 2024 research on quantized models, fine-tuning and compact optimization, followed by 2025 work including Marlin and low-precision training.
Sources and credits
Photo: people.csail.mit.edu; creator not stated. License: Not stated by source. Source: people.csail.mit.edu. Cropped to a square and resized. Ask us to remove or replace this photo.