QURE: AI-Assisted and Automatically Verified UDF Inlining

Summary: QURE uses LLMs to translate Python/Pandas UDFs to SQL and a formal equivalence verifier to confirm translations across diverse UDFs. Imperative constructs modeled in an intermediate verifier derive SQL-semantics conditions, achieving 88% equivalence (84% translations) with 23x/12x speedups. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 7121
Venue: SIGMOD
Year: 2025
Pagerank: 5.2643416e-05
Overall Rank: 9,456 | 35.13%
DOI: 10.1145/3709716

Incoming Non-self Citations Over Time

Authors

1. Tarique Siddiqui (Microsoft)
2. Arnd Christian König (Microsoft)
3. Jiashen Cao (Georgia Institute of Technology)
4. Cong Yan (Snowflake)
5. Shuvendu K. Lahiri (Microsoft)

BibTeX Citation

@inproceedings{siddiqui_sigmod25,
        title = {{QURE: AI-Assisted and Automatically Verified UDF Inlining}},
        author = {Siddiqui, Tarique and König, Arnd Christian and Cao, Jiashen and Yan, Cong and Lahiri, Shuvendu K.},
        series = {{SIGMOD} '25},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3709716},
        url = {https://dl.acm.org/doi/10.1145/3709716},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 2 of 2 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
10,232	EncoderForge: Generating Efficient SQL for Encoders in Machine Learning Inference Pipelines	2026	SIGMOD	5.093636e-05
10,541	Scalable GPU Acceleration of Scalar Functions in Analytical Databases: Compilation, Benchmarking, and Optimization	2026	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 14 of 14 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
894	Froid: Optimization of Imperative Programs in a Relational Database	2018	VLDB	0.00013367658
1,442	An Architecture for Compiling UDF-centric Workflows	2015	VLDB	0.00010778486
2,651	Magpie: Python at Speed and Scale using Cloud Backends	2021	CIDR	8.2918086e-05
3,257	One WITH RECURSIVE is Worth Many GOTOs	2021	SIGMOD	7.590651e-05
3,605	Computation Reuse in Analytics Job Service at Microsoft	2018	SIGMOD	7.2640711e-05
3,992	Aggify: Lifting the Curse of Cursor Loops using Custom Aggregates	2020	SIGMOD	6.9695338e-05
4,079	Flexible Rule-Based Decomposition and Metadata Independence in Modin: A Parallel Dataframe System	2022	VLDB	6.9188052e-05
4,180	BlackMagic: Automatic Inlining of Scalar UDFs into SQL Queries with Froid	2019	VLDB	6.8487143e-05
4,292	Putting Pandas in a Box	2021	CIDR	6.7784325e-05
5,107	Stage: Query Execution Time Prediction in Amazon Redshift	2024	SIGMOD	6.3623786e-05
5,684	Snakes on a Plan: Compiling Python Functions into Plain SQL Queries	2022	SIGMOD	6.1224825e-05
5,828	PL/SQL Without the PL	2020	SIGMOD	6.073983e-05
6,015	Dear User-Defined Functions, Inlining isn't working out so great for us. Let's try batching to make our relationship work. Sincerely, SQL	2024	CIDR	6.008272e-05
9,509	Versatile Optimization of UDF-heavy Data Flows with Sofa	2014	SIGMOD	5.2581466e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	4,180	BlackMagic: Automatic Inlining of Scalar UDFs into SQL Queries with Froid	2019	VLDB
2	6,267	Automated Validating and Fixing of Text-to-SQL Translation with Execution Consistency	2025	SIGMOD
3	6,314	YeSQL: "You extend SQL" with Rich and Highly Performant User-Defined Functions in Relational Databases	2022	VLDB
4	5,791	Accelerating Python UDFs in Vectorized Query Execution	2022	CIDR
5	11,120	Welding Natural Language Queries to Analytics IRs with LLMs	2024	CIDR
6	10,200	Automating Database-Native Function Code Synthesis with LLMs	2026	SIGMOD
7	10,740	UDFBench: A Tool for Benchmarking UDF Queries on SQL Engines	2025	SIGMOD
8	10,217	DeepEye-SQL: A Software-Engineering-Inspired Text-to-SQL Framework	2026	SIGMOD
9	8,096	Efficient Execution of User-Defined Functions in SQL Queries	2023	VLDB
10	10,139	Leveraging Query Optimizers to Verify the Soundness of LLM-based Query Rewrites for Real-World Workloads, and More!	2026	CIDR