Divya Mahajan
Assistant Professor
Designs computer architectures and large-scale infrastructure for machine learning training and data storage.
Divya Mahajan's lab at Georgia Tech works on distributed, cloud & edge systems and accelerators, chips & architecture. 19 papers since 2023. 4 PhD students. 5 alumni. Applications close .
At a glance
- papers, 2023–2026
- 5/ year
- at T1 venues
- 0%
- PhD students
- 4
- MS students
- 0
- alumni listed
- 5
Papers by yearfrom the lab homepage, 2023–2026
- 7 papers2026 · T1 0 · collecting
- 6 papers2025 · T1 0
- 4 papers2024 · T1 0
- 2 papers2023 · T1 0
T1 venuesOther
Research focuspaper titles that use each topic's words
- Distributed, cloud & edge systems4 papers
- Accelerators, chips & architecture1 paper
- learning training1 paper
- designs computer0 papers
- computer architectures0 papers
Paper list
From the lab homepage.
20267 papers
- IEEE International Symposium on High-Performance Computer Architecture (HPCA), FebruaryAn Adaptive Vector Index Partitioning Scheme for Low-Latency RAG Pipeline (opens in a new tab)
- —Compute-Communication Overlap Is Not Free: A Cross-Layer Characterization in GPU LLM Workloads
- Conference on Machine Learning and Systems (MLSys), MayEarthSight: Efficient Scheduling for Scalable Low-Latency Satellite Intelligence (opens in a new tab)
- Conference on Machine Learning and Systems (MLSys), MayFlashlight: PyTorch Compiler Extensions to Accelerate Attention Variants
- Conference on Machine Learning and Systems (MLSys), MayNEST: Network- and Memory-Aware Device Placement for Distributed Deep Learning (opens in a new tab)
- —NeuSight: Predicting the Performance of Future AI on Emerging GPUs
- Conference on Machine Learning and Systems (MLSys), MayStream2LLM: Overlap Context Streaming and Prefill for Reduced Time-to-First-Token (opens in a new tab)
20256 papers
- Conference on Neural Information Processing Systems (NeurIPS), DecemberCATransformers: Carbon Aware Transformers Through Joint Model-Hardware Optimization (opens in a new tab)
- IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), MayCharacterizing Compute-Communication Overlap in GPU-Accelerated Distributed Deep Learning: Performance and Power Implications (opens in a new tab)
- International Symposium on Microarchitecture (MICRO), OctoberCharacterizing the Efficiency of Distributed Training: A Power, Performance, and Thermal Perspective (opens in a new tab)
- International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), April , IEEE Micro Top Picks special issue of "most significant papers in computer architecture based on novelty and long-term impact"!, Featured in MediumForecasting GPU Performance for Deep Learning Training and Inference (opens in a new tab)
- Design Automation and Test in Europe (DATE), AprilGliding the Slipstream: Popularity-Based Embedding Skipping for Recommender Training (opens in a new tab)
- International Symposium on Microarchitecture (MICRO), OctoberPimba: A Processing-in-Memory Acceleration for Post-Transformer Large Language Model Serving (opens in a new tab)
20244 papers
- International Symposium on Computer Architecture (ISCA), JuneA Heterogeneous Acceleration Pipeline for Training Deep Recommendation Systems (opens in a new tab)
- Proceedings of the VLDB Endowment (VLDB), MayAccelerating String-key Learned Index Structures via Memoization-based Incremental Training (opens in a new tab)
- International Conference on Machine Learning (ICML), July , Featured in Conversation articleIntegrated Device Placement and Hardware Architecture Search (opens in a new tab)
- International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), AprilNeuPIMs: A NPU-PIM Heterogeneous Acceleration for Batched Inference of Large Language Model (opens in a new tab)
20232 papers
- NeurIPS WorkshopAd-Rec: Advanced Feature Interactions to Address Covariate-Shifts in Recommendation Networks (opens in a new tab)
- Conference on Neural Information Processing Systems (NeurIPS), DecemberFLuID: Mitigating Stragglers in Federated Learning using Invariant Dropout (opens in a new tab)
Where alumni went5 listed
Apply to Georgia TechFall 2027
Georgia Tech Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- optional
- English
- TOEFL/IELTS
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNSJTU
Jianguo Yao
- Accelerators, chips & architecture
- DETUM
Erika Covi
- Accelerators, chips & architecture
- CNZhejiang
Kejie Huang
- Accelerators, chips & architecture
- USUIUC
Nam Sung Kim
- Accelerators, chips & architecture
- USUW
Michael Bedford Taylor
- Accelerators, chips & architecture
- USUSC
Peter A. Beerel
- Accelerators, chips & architecture
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.
Georgia Techcloses