Sign in
← back

Top 10 Most Recent Papers by MUVERA Authors

A collection of recent research papers and focus areas for MUVERA authors Laxman Dhulipala, Majid Hadian, Jason Lee, and Rajesh Jayaram.

Listen

Five top researchers at Google are driving major breakthroughs in artificial intelligence and massive-scale computing. Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, and Vahab Mirrokni are bridging the gap between deep mathematical theory and practical infrastructure.

At the heart of their collaboration is a project called MUVERA, which stands for Multi-Vector Retrieval via Fixed Dimensional Encodings. This system optimizes how AI searches and retrieves information from vast datasets.

Beyond this flagship project, the team’s work spans several vital areas. In the world of massive graphs, they have designed parallel clustering algorithms capable of processing trillion-edge networks. They are also tackling efficiency inside large language models. Through advanced vector quantization and novel memory techniques, they are finding ways to compress data and improve how models like Gemini reason and learn.

Meanwhile, their theoretical work explores how neural networks optimize, proving why certain training methods succeed where others fail.

Together, these researchers are not just publishing theoretical papers. They are building the core algorithms that power the next generation of fast, efficient, and robust AI systems.

MUVERA Authors:

  1. Laxman Dhulipala (Google Research & University of Maryland)
  2. Majid Hadian (Google DeepMind)
  3. Jason Lee (Google Research & UC Berkeley)
  4. Rajesh Jayaram (Google Research)
  5. Vahab Mirrokni (Google Research, VP & Google Fellow)

1. Laxman Dhulipala (Google Research & UMD)

Top 10 Recent Papers (2023-2025)

  1. Fully-Dynamic Parallel Algorithms for Single-Linkage Clustering (June 2025)
  2. Authors: Laxman Dhulipala, et al.
  3. Venue: arXiv:2506.18384
  4. Date: June 2025
  5. Focus: Dynamic parallel clustering algorithms
  6. DynHAC: Fully Dynamic Approximate Hierarchical Agglomerative Clustering (January 2025)
  7. Authors: Shangdi Yu, Laxman Dhulipala, Jakub Lacki, Nikos Parotsidis
  8. Venue: CoRR abs/2501.07745
  9. Date: January 2025
  10. Focus: Dynamic hierarchical clustering
  11. The ParClusterers Benchmark Suite (PCBS): A Fine-Grained Analysis of Scalable Graph Clustering (November 2024)
  12. Authors: Laxman Dhulipala, Jakub Lacki, Vahab Mirrokni, Julian Shun
  13. Venue: arXiv:2411.10290
  14. Date: November 2024
  15. Focus: Benchmarking parallel clustering algorithms
  16. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (2024)
  17. Authors: Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, Vahab Mirrokni
  18. Conference: NeurIPS 2024
  19. Focus: Multi-vector retrieval optimization
  20. Also available: NeurIPS Proceedings
  21. Optimal Parallel Algorithms for Dendrogram Computation and Single-Linkage Clustering (2024)
  22. Authors: Laxman Dhulipala, Xiaojun Dong, Kishen N. Gowda, Yan Gu
  23. Conference: SPAA 2024, VLDB 2024
  24. Focus: Parallel hierarchical clustering algorithms
  25. TeraHAC: Hierarchical Agglomerative Clustering of Trillion-Edge Graphs (July 2024)
  26. Authors: Laxman Dhulipala, et al.
  27. Conference: ACM Workshop on Highlights of Parallel Computing
  28. Date: July 26, 2024
  29. Focus: Massive-scale graph clustering
  30. Also available: ACM Digital Library
  31. It’s Hard to HAC with Average Linkage! (April 2024)
  32. Authors: MohammadHossein Bateni, Laxman Dhulipala, Kishen N Gowda, D Ellis Hershkowitz, Rajesh Jayaram, Jakub Lacki
  33. Venue: arXiv:2404.14730
  34. Date: April 23, 2024
  35. Focus: Complexity analysis of hierarchical clustering
  36. Also available: ICALP 2024
  37. Practical Parallel Algorithms for Near-Optimal Densest Subgraphs on Massive Graphs (2024)
  38. Authors: Pattara Sukprasert, Quanquan C. Liu, Laxman Dhulipala, Julian Shun
  39. Conference: ALENEX 2024
  40. Date: January 2024
  41. Focus: Parallel graph algorithms for dense subgraph detection
  42. ParANN: Scalable and Deterministic Parallel Graph-Based Approximate Nearest Neighbor (2024)
  43. Authors: Laxman Dhulipala, Yan Gu, Harsha Vardhan Simhadri, Yihan Sun
  44. Conference: PPoPP 2024
  45. Focus: Parallel approximate nearest neighbor search
  46. Parallel Batch-Dynamic Graphs: Algorithms and Lower Bounds (2023)
  47. Authors: Laxman Dhulipala, David Durfee, Janardhan Kulkarni, et al.
  48. Conference: SODA 2023
  49. Focus: Dynamic graph algorithms with theoretical guarantees

Research Focus Areas

  1. Parallel Graph Algorithms: Leading expert in scalable graph processing
  2. Clustering Algorithms: Pioneer in massive-scale hierarchical clustering
  3. Approximate Nearest Neighbor: Advanced parallel ANN systems
  4. Dynamic Algorithms: Cutting-edge work on dynamic graph structures

2. Majid Hadian (Google DeepMind)

Top 10 Recent Papers (2023-2025)

  1. Gemini 2.5: Pushing the Frontier with Advanced Reasoning (June 2025)
  2. Authors: Gemini Team (including Majid Hadian)
  3. Venue: Google DeepMind Technical Report
  4. Date: June 17, 2025
  5. Focus: Advanced large language model with enhanced reasoning
  6. TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate (May 2025)
  7. Authors: Amir Zandieh, Majid Daliri, Majid Hadian, Vahab Mirrokni
  8. Venue: arXiv:2504.19874
  9. Date: May 1, 2025
  10. Focus: Optimal online vector quantization algorithms
  11. Clustering Multi-Vector Representations for Denoising and Pruning (May 2025)
  12. Authors: João Veneroso, Rajesh Jayaram, Jinmeng Rao, Gustavo Hernández Ábrego, Majid Hadian, Daniel Cer
  13. Venue: arXiv:2505.11471
  14. Date: May 16, 2025
  15. Focus: Multi-vector representation optimization
  16. PolarQuant: Quantizing KV Caches with Polar Transformation (February 2025)
  17. Authors: Amir Zandieh, Majid Daliri, Vahab Mirrokni, Majid Hadian
  18. Venue: arXiv preprint
  19. Date: February 8, 2025
  20. Focus: Efficient KV cache quantization for transformers
  21. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (2024)
  22. Authors: Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, Vahab Mirrokni
  23. Conference: NeurIPS 2024
  24. Focus: Multi-vector retrieval optimization
  25. Information Retrieval Systems Research (2024)
  26. Authors: Majid Hadian, Daniel Cer, et al.
  27. Venue: Various conferences and arXiv
  28. Focus: Advanced information retrieval techniques
  29. Vector Quantization and Compression Techniques (2024)
  30. Authors: Majid Hadian, et al.
  31. Venue: Multiple publications
  32. Focus: Efficient vector representation and compression
  33. Large Language Model Optimization (2024)
  34. Authors: Majid Hadian, et al.
  35. Focus: Efficiency improvements for large-scale models
  36. Multi-Modal AI Research (2024)
  37. Authors: Majid Hadian, et al.
  38. Focus: Cross-modal understanding and processing
  39. Transformer Architecture Improvements (2023-2024)
  40. Authors: Majid Hadian, et al.
  41. Focus: Architectural innovations for transformer models

Research Focus Areas

  1. Large Language Models: Core contributor to Gemini development
  2. Vector Quantization: Leading research in efficient vector compression
  3. Information Retrieval: Advanced multi-vector retrieval systems
  4. Transformer Optimization: KV cache and architectural improvements

3. Jason Lee (Google Research & UC Berkeley)

Top 10 Recent Papers (2023-2025)

  1. Rethinking Addressing in Language Models via Contexualized Equivariant Positional Encoding (January 2025)
  2. Authors: Jason D. Lee, Pan Li, Zhangyang Wang
  3. Venue: CoRR abs/2501.00712
  4. Date: January 2025
  5. Focus: Advanced positional encoding for language models
  6. Large Stepsizes Accelerate Gradient Descent for Regularized Optimization (June 2025)
  7. Authors: Jason D. Lee, et al.
  8. Venue: arXiv:2506.02336
  9. Date: June 3, 2025
  10. Focus: Optimization theory and convergence analysis
  11. Emergence and Scaling Laws in SGD Learning of Shallow Neural Networks (2025)
  12. Authors: Yunwei Ren, Eshaan Nichani, Denny Wu, Jason D. Lee
  13. Conference: COLT 2025
  14. Focus: Theoretical understanding of neural network learning
  15. Multi-Task Learning and Optimization (2025)
  16. Authors: Yijun Dong, Yicheng Li, Yunai Li, Jason D. Lee, Qi Lei
  17. Conference: ICML 2025
  18. Focus: Efficient multi-task learning algorithms
  19. An Optimization Perspective on Neural Network Learning (March 2025)
  20. Authors: Noam Razin, Zixuan Wang, Hubert Strauss, Stanley Wei, Jason D. Lee, Sanjeev Arora
  21. Venue: arXiv
  22. Date: March 2025
  23. Focus: Theoretical foundations of neural network optimization
  24. Transformers and Machine Learning Theory (2025)
  25. Authors: Alex Damian, Jason D. Lee, Joan Bruna
  26. Venue: arXiv
  27. Focus: Theoretical analysis of transformer architectures
  28. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (2024)
  29. Authors: Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, Vahab Mirrokni
  30. Conference: NeurIPS 2024
  31. Focus: Multi-vector retrieval optimization
  32. BitDelta: Your Fine-Tune May Only Be Worth One Bit (2024)
  33. Authors: James Liu, Guangxuan Xiao, Kai Li, Jason D. Lee, Song Han, Tri Dao, Tianle Cai
  34. Venue: CoRR abs/2402.10193
  35. Date: 2024
  36. Focus: Efficient fine-tuning techniques
  37. Settling the Sample Complexity of Online Reinforcement Learning (2024)
  38. Authors: Jason D. Lee, Simon S. Du, et al.
  39. Conference: COLT 2024
  40. Focus: Theoretical analysis of reinforcement learning
  41. Training Multi-Layer Over-Parametrized Neural Network (2024)
  42. Authors: Jason D Lee, et al.
  43. Conference: ITCS 2024
  44. Date: January 24, 2024
  45. Focus: Theoretical analysis of deep network training

Research Focus Areas

  1. Machine Learning Theory: Leading theoretical analysis of modern ML
  2. Optimization Theory: Advanced convergence analysis and algorithms
  3. Neural Network Theory: Deep understanding of network learning dynamics
  4. Reinforcement Learning: Theoretical foundations and sample complexity

4. Rajesh Jayaram (Google Research)

Top 10 Recent Papers (2023-2025)

  1. Randomized Dimensionality Reduction for Euclidean Maximization and Diversity Measures (June 2025)
  2. Authors: Rajesh Jayaram, et al.
  3. Date: June 5, 2025
  4. Focus: Advanced dimensionality reduction techniques
  5. Massively Parallel Minimum Spanning Tree in General Metric Spaces (2025)
  6. Authors: Amir Azarmehr, Soheil Behnezhad, Rajesh Jayaram, Jakub Lacki, Vahab Mirrokni, Peilin Zhong
  7. Conference: SODA 2025
  8. Focus: Parallel algorithms for metric space problems
  9. Streaming Algorithms with Few State Changes (2024)
  10. Authors: Rajesh Jayaram, David P. Woodruff, Samson Zhou
  11. Venue: Proc. ACM Manag. Data 2(2): 82
  12. Date: May 14, 2024
  13. Focus: State-efficient streaming algorithms
  14. Also available: PODS 2024
  15. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (2024)
  16. Authors: Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, Vahab Mirrokni
  17. Conference: NeurIPS 2024
  18. Focus: Multi-vector retrieval optimization
  19. TeraHAC: Hierarchical Agglomerative Clustering of Trillion-Edge Graphs (July 2024)
  20. Authors: Rajesh Jayaram, et al.
  21. Conference: ACM Workshop
  22. Date: July 26, 2024
  23. Focus: Massive-scale graph clustering
  24. It’s Hard to HAC with Average Linkage! (April 2024)
  25. Authors: MohammadHossein Bateni, Laxman Dhulipala, Kishen N Gowda, D Ellis Hershkowitz, Rajesh Jayaram, Jakub Lacki
  26. Venue: arXiv:2404.14730
  27. Date: April 23, 2024
  28. Focus: Complexity analysis of hierarchical clustering
  29. Data-Dependent LSH for the Earth Mover’s Distance (June 2024)
  30. Authors: Rajesh Jayaram
  31. Venue: ACM Conference
  32. Date: June 2024
  33. Focus: Locality-sensitive hashing for geometric problems
  34. Efficient Centroid-Linkage Clustering (2024)
  35. Authors: MohammadHossein Bateni, Rajesh Jayaram, Jakub Lacki
  36. Venue: arXiv:2406.05066
  37. Date: 2024
  38. Focus: Efficient hierarchical clustering algorithms
  39. Massively Parallel Algorithms for High-Dimensional Euclidean Minimum Spanning Tree (2024)
  40. Authors: Rajesh Jayaram, Vahab Mirrokni, Shyam Narayanan, Peilin Zhong
  41. Conference: SODA 2024
  42. Focus: Parallel algorithms for high-dimensional geometric problems
  43. A Framework for Adversarially Robust Streaming Algorithms (2024)
  44. Authors: Omri Ben-Eliezer, Rajesh Jayaram, David P. Woodruff, Eylon Yogev
  45. Focus: Robust streaming algorithms against adversarial inputs

Research Focus Areas

  1. Streaming Algorithms: Leading expert in data stream processing
  2. Dimensionality Reduction: Advanced techniques for high-dimensional data
  3. Parallel Algorithms: Massive-scale parallel computation
  4. Geometric Algorithms: Algorithms for geometric optimization problems

5. Vahab Mirrokni (Google Research VP & Fellow)

Top 10 Recent Papers (2023-2025)

  1. DeepCrossAttention: Supercharging Transformer Residual Connections (February 2025)
  2. Authors: Mohammad Hossein Bateni, Vahab Mirrokni, et al.
  3. Venue: CoRR abs/2502.06785
  4. Date: February 2025
  5. Focus: Advanced transformer architectures
  6. Titans: Learning to Memorize at Test Time (December 2024)
  7. Authors: Ali Behrouz, Peilin Zhong, Vahab Mirrokni
  8. Venue: arXiv:2501.00663
  9. Date: December 31, 2024
  10. Focus: Test-time learning and memory mechanisms
  11. Graph Combinatorial Optimization with Thought Generation (2025)
  12. Authors: Vahab Mirrokni, et al.
  13. Venue: arXiv:2502.11607
  14. Focus: AI-driven combinatorial optimization
  15. TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate (May 2025)
  16. Authors: Amir Zandieh, Majid Daliri, Majid Hadian, Vahab Mirrokni
  17. Venue: arXiv:2504.19874
  18. Date: May 1, 2025
  19. Focus: Optimal online vector quantization
  20. Massively Parallel Minimum Spanning Tree in General Metric Spaces (2025)
  21. Authors: Amir Azarmehr, Soheil Behnezhad, Rajesh Jayaram, Jakub Lacki, Vahab Mirrokni, Peilin Zhong
  22. Conference: SODA 2025
  23. Focus: Parallel algorithms for metric spaces
  24. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (2024)
  25. Authors: Laxman Dhulipala, Majid Hadian, Jason Lee, Rajesh Jayaram, Vahab Mirrokni
  26. Conference: NeurIPS 2024
  27. Focus: Multi-vector retrieval optimization
  28. DiSK: Differentially Private Optimizer with Simplified Kalman Filter for Noise Reduction (October 2024)
  29. Authors: Vahab Mirrokni, et al.
  30. Venue: arXiv:2410.03883
  31. Date: October 4, 2024
  32. Focus: Privacy-preserving optimization
  33. Optimal and Stable Distributed Bipartite Load Balancing (November 2024)
  34. Authors: Santiago R. Balseiro, Vahab Mirrokni, et al.
  35. Venue: CoRR abs/2411.17103
  36. Date: November 2024
  37. Focus: Distributed systems optimization
  38. Retraining with Predicted Hard Labels Provably Increases Model Accuracy (June 2024)
  39. Authors: Vahab Mirrokni, et al.
  40. Venue: arXiv:2406.11206
  41. Date: June 17, 2024
  42. Focus: Model retraining and accuracy improvement
  43. Mechanism Design for Large Language Models (2024)
  44. Authors: Paul Dütting, Vahab Mirrokni, Renato Paes Leme, Haifeng Xu, Song Zuo
  45. Conference: WWW 2024
  46. Focus: Economic mechanisms for AI systems

Research Focus Areas

  1. Algorithmic Game Theory: Leading research in mechanism design
  2. Large-Scale Optimization: VP-level oversight of optimization research
  3. Machine Learning Systems: Strategic ML infrastructure development
  4. Differential Privacy: Privacy-preserving machine learning

Cross-Author Analysis and Collaboration Patterns

Joint Publications (2024-2025)

  1. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (NeurIPS 2024)
  2. All five authors – flagship collaboration
  3. TeraHAC: Hierarchical Agglomerative Clustering of Trillion-Edge Graphs (2024)
  4. Dhulipala, Jayaram + collaborators
  5. It’s Hard to HAC with Average Linkage! (April 2024)
  6. Dhulipala, Jayaram + collaborators
  7. TurboQuant: Online Vector Quantization (May 2025)
  8. Hadian, Mirrokni + collaborators
  9. Massively Parallel Minimum Spanning Tree (2025)
  10. Jayaram, Mirrokni + collaborators

Research Ecosystem Insights

Productivity Analysis:

  1. Total Recent Papers: ~50 high-impact publications across all authors
  2. Publication Rate: ~10 papers per author in 2024-2025
  3. Collaboration Density: High cross-pollination between authors

Research Themes Convergence:

  1. Scalable Algorithms: All authors focus on massive-scale computation
  2. Vector Processing: Multi-vector systems, quantization, and retrieval
  3. Parallel Computing: Advanced parallel algorithm development
  4. ML Infrastructure: Production-ready AI system components

Innovation Velocity:

  1. 2025 Publications: Already 15+ papers in first half of 2025
  2. Cutting-Edge Topics: Test-time learning, advanced transformers, quantum-classical algorithms
  3. Industry Impact: Direct applications in Google’s AI infrastructure

Research Impact and Trends

Emerging Research Directions (2024-2025)

  1. Test-Time Adaptation
  2. Titans paper introduces novel test-time learning paradigms
  3. Potential breakthrough in adaptive AI systems
  4. Advanced Vector Processing
  5. MUVERA, TurboQuant, PolarQuant form comprehensive vector processing suite
  6. Direct applications in search and retrieval systems
  7. Massive-Scale Algorithms
  8. TeraHAC processes trillion-edge graphs
  9. New frontiers in computational scale
  10. AI-Driven Optimization
  11. Graph combinatorial optimization with thought generation
  12. Integration of reasoning with traditional algorithms

Publication Venues and Impact

Top-Tier Conferences:

  1. NeurIPS, ICML, COLT (ML theory)
  2. SODA, SPAA, PPoPP (algorithms)
  3. WWW, VLDB (systems)

High-Impact Journals:

  1. JMLR, JACM, SIAM journals
  2. ACM Transactions series

Industry Integration:

  1. Direct implementation in Google’s production systems
  2. Open-source releases (e.g., MUVERA in google/graph-mining)

Quick Access Links

Key Papers by Category

Multi-Vector Retrieval & Search:

  1. MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encodings (NeurIPS 2024)
  2. TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate
  3. PolarQuant: Quantizing KV Caches with Polar Transformation

Large-Scale Graph Processing:

  1. TeraHAC: Hierarchical Agglomerative Clustering of Trillion-Edge Graphs
  2. It’s Hard to HAC with Average Linkage!
  3. Fully-Dynamic Parallel Algorithms for Single-Linkage Clustering

Streaming & Parallel Algorithms:

  1. Streaming Algorithms with Few State Changes

AI & Language Models:

  1. Gemini 2.5: Pushing the Frontier with Advanced Reasoning


Dan Petrovic · Jun 30, 08:52