Parth Suresh

I’m a Member of Technical Staff at Datology AI, where I work on synthetic data generation for web-scale and long-context data.

Previously, I was a Software Engineer (Machine Learning) at Meta Reality Labs, working on synthetic data, data curation, and benchmarks for language and multimodal models, including in egocentric and wearable settings. Before that I was an ML Research Engineer at Scale AI, focused on reasoning, synthetic data generation, and LLM judges. Earlier at Meta I worked at the intersection of developer productivity, software engineering, and machine learning.

The question I keep coming back to is how we turn messy, real-world data into reliable signals for training and evaluating large models.

News

Aug 10, 2026	CRAG-MM won the Best Paper Award at KDD 2026 in the Datasets and Benchmarks track!
Jul 20, 2026	Joined Datology AI as a Member of Technical Staff, working on synthetic data generation for web and long-context data.
Jun 03, 2026	Our paper Plan, Watch, Recover, a benchmark and architectures for proactive procedural assistance, is now on arXiv.
Oct 30, 2025	Released CRAG-MM, a multi-modal multi-turn RAG benchmark for wearable / egocentric settings - and the foundation for KDD Cup 2025.
Dec 10, 2024	Our paper on balancing cost and effectiveness of synthetic data generation strategies for LLMs was accepted to the FITML Workshop at NeurIPS 2024.

Selected publications

CRAG-MM: Multi-modal Multi-turn Comprehensive RAG Benchmark

Jiaqi Wang, Xiao Yang, Kai Sun, Parth Suresh, Sanat Sharma, and 36 more authors

In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2026

Best Paper Bib PDF

Best Paper Award at KDD 2026 in the Datasets and Benchmarks track

@inproceedings{wang2025cragmm,
  title = {{CRAG-MM}: Multi-modal Multi-turn Comprehensive {RAG} Benchmark},
  author = {Wang, Jiaqi and Yang, Xiao and Sun, Kai and Suresh, Parth and Sharma, Sanat and Czyzewski, Adam and Andersen, Derek and Appini, Surya and Banerjee, Arkav and Choudhary, Sajal and Ghasemlou, Shervin and Guan, Ziqiang and Iyer, Akil and Khan, Haidar and Kong, Lingkun and Luo, Roy and Ma, Tiffany and Qiao, Zhen and Tran, David and Xu, Wenfang and Yeatman, Skyler and Zhou, Chen and Gujral, Gunveer and Xia, Yinglong and Moon, Shane and Scheffer, Nicolas and Shah, Nirav and Chang, Eun and Liu, Yue and Metze, Florian and Stark, Tammy and Feizollahi, Zhaleh and Jessee, Andrea and Pujari, Mangesh and Aly, Ahmed and Damavandi, Babak and Wanga, Rakesh and Kumar, Anuj and Patel, Rohit and Yih, Wen-tau and Dong, Xin Luna},
  booktitle = {Proceedings of the {ACM} {SIGKDD} Conference on Knowledge Discovery and Data Mining ({KDD})},
  year = {2026},
}

Improving Code Reviewer Recommendation: Accuracy, Latency, Workload, and Bystanders

Peter C. Rigby, Seth Rogers, Sadruddin Saleem, Parth Suresh, Daniel Suskin, and 4 more authors

ACM Transactions on Software Engineering and Methodology, 2025

Bib PDF

@article{rigby2025code,
  title = {Improving Code Reviewer Recommendation: Accuracy, Latency, Workload, and Bystanders},
  author = {Rigby, Peter C. and Rogers, Seth and Saleem, Sadruddin and Suresh, Parth and Suskin, Daniel and Riggs, Patrick and Maddila, Chandra and Nagappan, Nachiappan and Mockus, Audris},
  journal = {ACM Transactions on Software Engineering and Methodology},
  volume = {35},
  number = {1},
  pages = {1--23},
  year = {2025},
}

Balancing Cost and Effectiveness of Synthetic Data Generation Strategies for LLMs

Yung-Chieh Chan, George Pu, Apaar Shanker, Parth Suresh, Penn Jenks, and 2 more authors

In Fine-Tuning in Machine Learning (FITML) Workshop at NeurIPS, 2024

Bib PDF

@inproceedings{chan2024balancing,
  title = {Balancing Cost and Effectiveness of Synthetic Data Generation Strategies for {LLMs}},
  author = {Chan, Yung-Chieh and Pu, George and Shanker, Apaar and Suresh, Parth and Jenks, Penn and Heyer, John and Denton, Sam},
  booktitle = {Fine-Tuning in Machine Learning ({FITML}) Workshop at {NeurIPS}},
  year = {2024},
}