Andrew Drozdov

About me

I’m a research scientist at Databricks working on scaling intelligence through search and for search. I design and train search methods that integrate with agents, helping them find information, reason across sources, and solve complex tasks. Recent projects include KARL, Instructed Retriever, AutoIndex, and FreshStack.

I received my PhD from UMass Amherst CICS, advised by Andrew McCallum and Mohit Iyyer, and my MS in computer science from NYU, where I worked with Samuel Bowman and Kyunghyun Cho. Previously, I worked at Google and IBM.

I’ve reviewed more than 100 papers at AI, information retrieval, and NLP conferences and served as an area chair and senior area chair.

Say hello: andrew.drozdov@databricks.com

Updates

September 11, 2026	Virginia Tech Frontier AI Seminar, hosted by Tu Vu: “Three Reasons You Should Train Search Agents.”
July 24, 2026	SIGIR LLMUP Workshop keynote: “Training Adaptive Search Agents for Dynamic Environments.”
June 4, 2026	Databricks blog: “3x Faster Search: Parallel Test-Time Scaling with Instructed-Retriever-1.”
April 22, 2026	NYU Fundamentals of Machine Learning guest lecture, hosted by Kyunghyun Cho: “Grounded Reasoning in Real-World ML Systems.”
January 6, 2026	Databricks blog: “Instructed Retriever: Unlocking System-Level Reasoning in Search Agents.”

All updates →

Publications

AutoIndex: Learning Representation Programs for Retrieval

Sam O’Nuallain, Nithya Rajkumar, Ramya Narayanasamy, Hanna Jiang, Shreyas Chaudhari, and Andrew Drozdov

arXiv 2026

arXiv Bib
@article{onuallain2026autoindex, bibtex_show = {true}, selected = {true}, category = {main}, title = {{AutoIndex}: Learning Representation Programs for Retrieval}, author = {O'Nuallain, Sam and Rajkumar, Nithya and Narayanasamy, Ramya and Jiang, Hanna and Chaudhari, Shreyas and Drozdov, Andrew}, journal = {arXiv}, year = {2026}, arxiv = {https://arxiv.org/abs/2607.18603} }

Can QPP Choose the Right Query Variant? Evaluating Query Variant Selection for RAG Pipelines

Negar Arabzadeh, Andrew Drozdov, Michael Bendersky, and Matei Zaharia

In SIGIR 2026

arXiv Bib HTML

@inproceedings{arabzadeh2026qpp,
  bibtex_show = {true},
  selected = {true},
  category = {main},
  title = {Can {QPP} Choose the Right Query Variant? Evaluating Query Variant Selection for {RAG} Pipelines},
  author = {Arabzadeh, Negar and Drozdov, Andrew and Bendersky, Michael and Zaharia, Matei},
  booktitle = {SIGIR},
  year = {2026},
  pages = {2813--2823},
  doi = {10.1145/3805712.3808571},
  arxiv = {https://arxiv.org/abs/2604.22661},
  html = {https://doi.org/10.1145/3805712.3808571}
}

KARL: Knowledge Agents via Reinforcement Learning

Jonathan D. Chang, Andrew Drozdov, Shubham Toshniwal, Owen Oertell, Alexander Trott, Jacob Portes, Abhay Gupta, Pallavi Koppol, Ashutosh Baheti, Sean Kulinski, Ivan Zhou, Irene Dea, Krista Opsahl-Ong, Simon Favreau-Lessard, Sean Owen, Jose Javier Gonzalez Ortiz, Arnav Singhvi, Xabi Andrade, Cindy Wang, Kartik Sreenivasan, Sam Havens, Jialu Liu, Peyton DeNiro, Wen Sun, Michael Bendersky, and Jonathan Frankle

arXiv 2026

arXiv Bib
@article{chang2026karl, bibtex_show = {true}, selected = {true}, category = {main}, title = {{KARL}: Knowledge Agents via Reinforcement Learning}, author = {Chang, Jonathan D. and Drozdov, Andrew and Toshniwal, Shubham and Oertell, Owen and Trott, Alexander and Portes, Jacob and Gupta, Abhay and Koppol, Pallavi and Baheti, Ashutosh and Kulinski, Sean and Zhou, Ivan and Dea, Irene and Opsahl-Ong, Krista and Favreau-Lessard, Simon and Owen, Sean and Gonzalez Ortiz, Jose Javier and Singhvi, Arnav and Andrade, Xabi and Wang, Cindy and Sreenivasan, Kartik and Havens, Sam and Liu, Jialu and DeNiro, Peyton and Sun, Wen and Bendersky, Michael and Frankle, Jonathan}, journal = {arXiv}, year = {2026}, arxiv = {https://arxiv.org/abs/2603.05218} }
A State-of-the-Art SQL Reasoning Model using RLVR

Alnur Ali, Ashutosh Baheti, Jonathan Chang, Ta-Chung Chi, Brandon Cui, Andrew Drozdov, Jonathan Frankle, Abhay Gupta, Pallavi Koppol, Sean Kulinski, Jonathan Li, Dipendra Misra, Krista Opsahl-Ong, Jose Javier Gonzalez Ortiz, Matei Zaharia, and Yue Zhang

arXiv 2025

arXiv Bib
@article{ali2025sqlrlvr, bibtex_show = {true}, selected = {true}, category = {main}, title = {A State-of-the-Art {SQL} Reasoning Model using {RLVR}}, author = {Ali, Alnur and Baheti, Ashutosh and Chang, Jonathan and Chi, Ta-Chung and Cui, Brandon and Drozdov, Andrew and Frankle, Jonathan and Gupta, Abhay and Koppol, Pallavi and Kulinski, Sean and Li, Jonathan and Misra, Dipendra and Opsahl-Ong, Krista and Gonzalez Ortiz, Jose Javier and Zaharia, Matei and Zhang, Yue}, journal = {arXiv}, year = {2025}, arxiv = {https://arxiv.org/abs/2509.21459} }
FreshStack: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents

Nandan Thakur, Jimmy Lin, Sam Havens, Michael Carbin, Omar Khattab, and Andrew Drozdov

arXiv 2025

arXiv Bib
@article{thakur2025freshstack, bibtex_show = {true}, selected = {true}, category = {main}, title = {{FreshStack}: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents}, author = {Thakur, Nandan and Lin, Jimmy and Havens, Sam and Carbin, Michael and Khattab, Omar and Drozdov, Andrew}, journal = {arXiv}, year = {2025}, arxiv = {https://arxiv.org/abs/2504.13128} }
The Second Tutorial on Retrieval-Enhanced Machine Learning: Synthesis and Opportunities

Fernando Diaz, Andrew Drozdov, To Eun Kim, Alireza Salemi, and Hamed Zamani

In SIGIR 2025

Bib HTML
@inproceedings{diaz2025remal, bibtex_show = {true}, selected = {true}, category = {main}, abbr = {Tutorial}, title = {The Second Tutorial on {Retrieval-Enhanced Machine Learning}: Synthesis and Opportunities}, author = {Diaz, Fernando and Drozdov, Andrew and Kim, To Eun and Salemi, Alireza and Zamani, Hamed}, booktitle = {SIGIR}, year = {2025}, pages = {4130--4133}, doi = {10.1145/3726302.3731695}, html = {https://doi.org/10.1145/3726302.3731695} }
Drowning in Documents: Consequences of Scaling Reranker Inference

Mathew Jacob, Erik Lindgren, Matei Zaharia, Michael Carbin, Omar Khattab, and Andrew Drozdov

arXiv 2024

arXiv Bib
@article{jacobs2024drowning, bibtex_show = {true}, selected = {true}, category = {main}, title = {{Drowning in Documents}: Consequences of Scaling Reranker Inference}, author = {Jacob, Mathew and Lindgren, Erik and Zaharia, Matei and Carbin, Michael and Khattab, Omar and Drozdov, Andrew}, journal = {arXiv}, year = {2024}, arxiv = {https://arxiv.org/abs/2411.11767} }
Retrieval-Enhanced Machine Learning: Synthesis and Opportunities

Fernando Diaz, Andrew Drozdov, To Eun Kim, Alireza Salemi, and Hamed Zamani

SIGIR-AP 2024

Bib HTML
@article{diaz2024remal, bibtex_show = {true}, selected = {true}, category = {main}, abbr = {Tutorial}, title = {{Retrieval-Enhanced Machine Learning}: Synthesis and Opportunities}, author = {Diaz, Fernando and Drozdov, Andrew and Kim, To Eun and Salemi, Alireza and Zamani, Hamed}, journal = {SIGIR-AP}, year = {2024}, html = {https://dl.acm.org/doi/abs/10.1145/3673791.3698439} }

Unlocking Natural Language Generalization with Adaptive Retrieval-based Methods

Andrew Drozdov

2024

Bib HTML

@article{drozdov2024thesis,
  bibtex_show = {true},
  selected = {true},
  category = {main},
  abbr = {Thesis},
  title = {Unlocking Natural Language Generalization with Adaptive Retrieval-based Methods},
  author = {Drozdov, Andrew},
  year = {2024},
  html = {https://scholarworks.umass.edu/entities/publication/04520d37-6227-4441-9ec8-e91495271ce8}
}

ReDMM: Retrieval Driven Memory Manager

Andrew Drozdov, Andrew McCallum, and Mohit Iyyer

2024

Bib
@article{drozdov2024redmm, bibtex_show = {true}, selected = {true}, category = {main}, title = {{ReDMM}: Retrieval Driven Memory Manager}, author = {Drozdov, Andrew and McCallum, Andrew and Iyyer, Mohit}, year = {2024} }
Multistage collaborative knowledge distillation from a large language model for semi-supervised sequence generation

Jiachen Zhao, Wenlong Zhao, Andrew Drozdov, Benjamin Rozonoyer, Md Arafat Sultan, Jay Yoon Lee, Mohit Iyyer, and Andrew McCallum

In ACL 2024

arXiv Bib
@inproceedings{zhao2024multistage, bibtex_show = {true}, selected = {true}, category = {main}, abbr = {Long Paper}, title = {Multistage collaborative knowledge distillation from a large language model for semi-supervised sequence generation}, author = {Zhao, Jiachen and Zhao, Wenlong and Drozdov, Andrew and Rozonoyer, Benjamin and Sultan, Md Arafat and Lee, Jay Yoon and Iyyer, Mohit and McCallum, Andrew}, booktitle = {ACL}, year = {2024}, arxiv = {https://aclanthology.org/2024.acl-long.766/} }
PaRaDe: Passage Ranking using Demonstrations with LLMs

Andrew Drozdov, Honglei Zhuang, Zhuyun Dai, Zhen Qin, Razieh Rahimi, Xuanhui Wang, Dana Alon, Mohit Iyyer, Andrew McCallum, Donald Metzler, and Kai Hui

In EMNLP (Findings) 2023

arXiv Bib
@inproceedings{drozdov2023parade, bibtex_show = {true}, selected = {true}, abbr = {Short Paper}, category = {main}, title = {{PaRaDe}: Passage Ranking using Demonstrations with LLMs}, author = {Drozdov, Andrew and Zhuang, Honglei and Dai, Zhuyun and Qin, Zhen and Rahimi, Razieh and Wang, Xuanhui and Alon, Dana and Iyyer, Mohit and McCallum, Andrew and Metzler, Donald and Hui, Kai}, year = {2023}, booktitle = {EMNLP (Findings)}, arxiv = {https://arxiv.org/abs/2310.14408} }
kNN-LM Does Not Improve Open-ended Text Generation

Shufan Wang, Yixiao Song, Andrew Drozdov, Aparna Garimella, Varun Manjunatha, and Mohit Iyyer

In EMNLP 2023

arXiv Bib
@inproceedings{wang2023knnlm, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, category = {main}, title = {{kNN-LM} Does Not Improve Open-ended Text Generation}, author = {Wang, Shufan and Song, Yixiao and Drozdov, Andrew and Garimella, Aparna and Manjunatha, Varun and Iyyer, Mohit}, year = {2023}, booktitle = {EMNLP}, arxiv = {https://arxiv.org/abs/2305.14625} }
Compositional Semantic Parsing with Large Language Models

Andrew Drozdov, Nathanael Schärli, Ekin Akyürek, Nathan Scales, Xinying Song, Xinyun Chen, Olivier Bousquet, and Denny Zhou

In ICLR 2023

arXiv Bib
@inproceedings{drozdov2022compositional, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, category = {main}, title = {Compositional Semantic Parsing with Large Language Models}, author = {Drozdov, Andrew and Schärli, Nathanael and Akyürek, Ekin and Scales, Nathan and Song, Xinying and Chen, Xinyun and Bousquet, Olivier and Zhou, Denny}, year = {2023}, booktitle = {ICLR}, award = {Poster}, arxiv = {https://arxiv.org/abs/2209.15003} }
You can’t pick your neighbors, or can you? When and how to rely on retrieval in the kNN-LM

Andrew Drozdov, Shufan Wang, Razieh Rahimi, Andrew McCallum, Hamed Zamani, and Mohit Iyyer

In EMNLP (Findings) 2022

Bib PDF
@inproceedings{drozdov2022knnlm, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, category = {main}, title = {You can't pick your neighbors, or can you? {W}hen and how to rely on retrieval in the {kNN-LM}}, author = {Drozdov, Andrew and Wang, Shufan and Rahimi, Razieh and McCallum, Andrew and Zamani, Hamed and Iyyer, Mohit}, booktitle = {EMNLP (Findings)}, award = {Findings}, year = {2022}, pdf = {https://mrdrozdov.github.io/knnlm_retrieval_quality.pdf} }
Inducing and Using Alignments for Transition-based AMR Parsing

Andrew Drozdov, Jiawei Zhou, Radu Florian, Andrew McCallum, Tahira Naseem, Yoon Kim, and Ramon Fernandez Astudillo

In NAACL 2022

arXiv Bib Code
@inproceedings{drozdov2022amralign, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, award = {Poster}, category = {main}, title = {Inducing and Using Alignments for Transition-based AMR Parsing}, author = {Drozdov, Andrew and Zhou, Jiawei and Florian, Radu and McCallum, Andrew and Naseem, Tahira and Kim, Yoon and Astudillo, Ramon Fernandez}, booktitle = {NAACL}, year = {2022}, arxiv = {https://arxiv.org/abs/2205.01464}, code = {https://github.com/IBM/transition-amr-parser} }
Improved Latent Tree Induction with Distant Supervision via Span Constraints

Zhiyang Xu, Andrew Drozdov, Jay Yoon Lee, Tim O’Gorman, Subendhu Rongali, Dylan Finkbeiner, Shilpa Suresh, Mohit Iyyer, and Andrew McCallum

In EMNLP 2021

arXiv Bib Code
@inproceedings{diora2021distantdiora, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, award = {Poster}, category = {main}, title = {Improved Latent Tree Induction with Distant Supervision via Span Constraints}, author = {Xu, Zhiyang and Drozdov, Andrew and Lee, Jay Yoon and O{'}Gorman, Tim and Rongali, Subendhu and Finkbeiner, Dylan and Suresh, Shilpa and Iyyer, Mohit and McCallum, Andrew}, booktitle = {EMNLP}, year = {2021}, arxiv = {https://arxiv.org/abs/2109.05112}, code = {https://github.com/iesl/distantly-supervised-diora} }
Unsupervised Parsing with S-DIORA: Single Tree Encoding for Deep Inside-Outside Recursive Autoencoders

Andrew Drozdov, Subendhu Rongali, Yi-Pei Chen, Tim O’Gorman, Mohit Iyyer, and Andrew McCallum

In EMNLP 2020

Bib PDF
@inproceedings{drozdov2020sdiora, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, award = {Poster}, category = {main}, title = {Unsupervised Parsing with {S-DIORA}: Single Tree Encoding for Deep Inside-Outside Recursive Autoencoders}, author = {Drozdov, Andrew and Rongali, Subendhu and Chen, Yi-Pei and O{'}Gorman, Tim and Iyyer, Mohit and McCallum, Andrew}, booktitle = {EMNLP}, pdf = {https://aclanthology.org/2020.emnlp-main.392/}, year = {2020} }

The Impact of Preprint Servers in the Formation of Novel Ideas

Swarup Satish, Zonghai Yao, Andrew Drozdov, and Boris Veytsman

In EMNLP (Workshop on Scholarly Document Processing) 2020

arXiv Bib Code Workshop

@inproceedings{emnl2020biorxivimpact,
  bibtex_show = {true},
  selected = {true},
  category = {workshop},
  abbr = {Long Paper},
  award = {Oral},
  arxiv = {https://www.biorxiv.org/content/10.1101/2020.10.08.330696v1},
  code = {https://github.com/seasonyao/BiorXivImpact},
  title = {The Impact of Preprint Servers in the Formation of Novel Ideas},
  author = {Satish, Swarup and Yao, Zonghai and Drozdov, Andrew and Veytsman, Boris},
  booktitle = {EMNLP (Workshop on Scholarly Document Processing)},
  workshop = {https://ornlcda.github.io/SDProc/},
  year = {2020}
}

Unsupervised Labeled Parsing with Deep Inside-Outside Recursive Auto-Encoders

Andrew Drozdov, Patrick Verga, Yi-Pei Chen, Mohit Iyyer, and Andrew McCallum

In EMNLP 2019

Bib PDF Code
@inproceedings{drozdov2019dioralabeled, bibtex_show = {true}, selected = {true}, abbr = {Short Paper}, award = {Poster}, category = {main}, author = {Drozdov, Andrew and Verga, Patrick and Chen, Yi-Pei and Iyyer, Mohit and McCallum, Andrew}, title = {Unsupervised Labeled Parsing with Deep Inside-Outside Recursive Auto-Encoders}, booktitle = {EMNLP}, pdf = {https://aclanthology.org/D19-1161/}, code = {https://github.com/iesl/diora}, year = {2019} }
Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders

Andrew Drozdov, Patrick Verga, Mohit Yadav, Mohit Iyyer, and Andrew McCallum

In NAACL 2019

arXiv Bib Code
@inproceedings{drozdov2019diora, bibtex_show = {true}, selected = {true}, abbr = {Long Paper}, award = {Oral}, category = {main}, author = {Drozdov, Andrew and Verga, Patrick and Yadav, Mohit and Iyyer, Mohit and McCallum, Andrew}, title = {Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders}, booktitle = {NAACL}, arxiv = {https://arxiv.org/abs/1904.02142}, code = {https://github.com/iesl/diora}, year = {2019} }

Do latent tree learning models identify meaningful structure in sentences?

Adina Williams, Andrew Drozdov, and Samuel R. Bowman

TACL 2018

arXiv Bib Code

@article{Williams2018DoLT,
  bibtex_show = {true},
  selected = {true},
  category = {main},
  abbr = {Journal},
  award = {Oral},
  title = {Do latent tree learning models identify meaningful structure in sentences?},
  author = {Williams, Adina and Drozdov, Andrew and Bowman, Samuel R.},
  journal = {TACL},
  year = {2018},
  volume = {6},
  code = {https://github.com/nyu-mll/spinn},
  arxiv = {https://arxiv.org/abs/1709.01121},
  pages = {253-267}
}

Emergent Communication in a Multi-Modal, Multi-Step Referential Game

Katrina Evtimova, Andrew Drozdov, Douwe Kiela, and Kyunghyun Cho

In ICLR 2018

arXiv Bib Code

@inproceedings{iclr2018emergent,
  bibtex_show = {true},
  selected = {true},
  category = {main},
  abbr = {Long Paper},
  award = {Poster},
  arxiv = {https://arxiv.org/abs/1705.10369},
  title = {Emergent Communication in a Multi-Modal, Multi-Step Referential Game},
  author = {Evtimova, Katrina and Drozdov, Andrew and Kiela, Douwe and Cho, Kyunghyun},
  booktitle = {ICLR},
  year = {2018},
  code = {https://github.com/nyu-dl/MultimodalGame}
}

The Coadaptation Problem when Learning How and What to Compose

Andrew Drozdov, and Samuel R. Bowman

In ACL (Workshop on Representation Learning for NLP) 2017

Bib PDF Workshop

@inproceedings{acl2017coadapt,
  bibtex_show = {true},
  selected = {true},
  category = {workshop},
  abbr = {Ext. Abstract},
  award = {Poster},
  pdf = {https://github.com/nyu-mll/spinn/blob/master/writing/acl17-latex/acl2017.pdf},
  title = {The Coadaptation Problem when Learning How and What to Compose},
  author = {Drozdov, Andrew and Bowman, Samuel R.},
  booktitle = {ACL (Workshop on Representation Learning for NLP)},
  workshop = {https://sites.google.com/site/repl4nlp2017/},
  year = {2017}
}