Abhinav Gorantla
I am a Ph.D. student in Computer Science at Arizona State University, working with Prof. K. Selçuk Candan in EMIT Lab. Broadly, I study how ideas from causality can improve machine learning. During my M.S. at ASU (completed with distinction), I contributed to CausalBench, an open benchmarking platform for causal ML (Best Demo Award, CIKM 2024; tutorial at KDD 2025), and co-authored work on causally-informed data retrieval at SIGMOD 2026. I’m currently investigating how causal structure can be leveraged to improve the robustness and generalization of ML models.
Research
My research lies at the intersection of causal machine learning and multi-objective optimization. I am especially interested in:
- Causal structure for ML generalization. Investigating how causal priors can improve the robustness and generalization of ML models.
- Causal benchmarking and systems. Designing and maintaining CausalBench, a benchmarking framework and service for causal learning algorithms (causal discovery, causal inference, and interpretability).
- Multi-objective optimization & skyline queries. Studying efficient Skyline/Pareto-style retrieval in relational data, including how causal structure can impact efficiency of Pareto-style retrieval algorithms.
Education
- Doctor of Philosophy in Computer Science, Arizona State University, Jan 2026 - Present
- GPA: 4.00/4.00
- Advisor: Prof. K. Selçuk Candan
- Master of Science in Computer Science, Arizona State University, Dec 2025
- GPA: 4.00/4.00
- Honors: with Distinction
- Advisor: Prof. K. Selçuk Candan
- Bachelor of Technology in Computer Science and Engineering, Vellore Institute of Technology, May 2023
- GPA: 8.94/10.00
Publications
IEEE BigData '26 VoS: Variate Ordering Strategies for Skyline Query Optimization
Accepted to the 2026 IEEE International Conference on Big Data (IEEE BigData 2026)
Skyline query efficiency depends heavily on data characteristics, and in this paper we argue that algorithms should optimize per-attribute dominance checks (not just per-tuple ones), proposing variate-ordering strategies to cut redundant attribute comparisons.
CIKM '26 The Good, the Bad, and the Ugly of Markov Boundary for Tabular Prediction
Accepted to the 35th ACM International Conference on Information and Knowledge Management (CIKM ’26)
Showed that a target variable oracle Markov boundary can substantially improve tabular prediction, but recovering it via causal discovery fails to deliver — motivating prediction-aligned feature selection.
SIGMOD '26 Causal Search for Skylines (CSS): Causally-Informed Selective Data De-Correlation
Proceedings of the ACM on Management of Data, Volume 4, Issue 3 (SIGMOD 2026)
Proposes a method for leveraging causal structure to selectively de-correlate data, improving efficiency of Skyline retrieval in relational databases.
WSDM '26 CausalBench+: Causal-Informed Machine Learning Benchmarking
Proceedings of the Nineteenth ACM International Conference on Web Search and Data Mining (WSDM Companion '26), Boise, ID, USA, 2026, pp. 120–122
Extends CausalBench beyond the causal learning domain, enabling fair benchmarking across general machine learning tasks through user-provided task definitions.
CIKM '25 CausalBench-ER: Causally-Informed Explanations and Recommendations for Reproducible Benchmarking
Proceedings of the 34th ACM International Conference on Information and Knowledge Management (CIKM ’25), Seoul, Republic of Korea, 2025, pp. 6426–6431
Extends CausalBench with causally-informed explanations and recommendations to help researchers better understand and reproduce their results.
CIKM '24Best Demo Award Introducing CausalBench: A Flexible Benchmark Framework for Causal Analysis and Machine Learning
Proceedings of the 33rd ACM International Conference on Information and Knowledge Management (CIKM 2024), pp. 5220–5224
Introduces an open platform that standardizes how causal learning algorithms are evaluated and compared across the research community.
Tutorials
SIGSPATIAL '26 Spatio-Causal Modeling and Applications
Accepted to the 34th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2026), Riverside, California, USA
Introduces causal and spatio-causal foundations, surveys algorithms for four major spatial decision-support tasks, and concludes with hands-on spatio-causal modeling through CausalBench.
KDD '25 CausalBench: Causal Learning Research Streamlined
Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’25), Vol. 2, Toronto, Canada, 2025, pp. 6239–6240
Overview of causal machine learning, challenges in benchmarking, and hands-on use of the CausalBench platform.
Professional Service
- Program Committee Member, ACM Conference on Information and Knowledge Management (CIKM 2026)
- Student Volunteer, ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025)
Experience
- August 2024 - Now: Graduate Research Assistant at Emitlab, School of Computing and Augmented Intelligence, ASU
- Working on causally informed training methods for Prior Fitted Networks (PFNs).
- Developing an optimized algorithm for efficient Skyline retrieval in relational database systems by leveraging causal structure in the data.
- Conducting research on optimizing the selection process in multi-objective evolutionary algorithms, specifically Pareto-based methods like NSGA and SPEA.
- Collaborating with researchers at CASCADE Lab to maintain and improve causalbench.org, a platform dedicated to causal discovery benchmarks.
- Contributed to the engineering effort on APPEX’s document-ingestion pipeline, which converts user-uploaded documents into a queryable causal knowledge graph, as part of the Meta-Annotation Team.
- Built a causal-analysis recommendation system; led end-to-end integration and serverless deployment on AWS Lambda; published at CIKM 2025.
- March 2024 – August 2024: Graduate Services Assistant at Emitlab, School of Computing and Augmented Intelligence, ASU
- Helped develop the
causalbenchPython package and causalbench.org website as an end-to-end benchmarking solution for the causal machine learning community. - Served as a full-stack developer on CausalBench, integrating datasets, models, and metrics into a unified evaluation workflow.
- Redesigned the backend architecture for the Skysong project, enabling responsive causal analyses in production (≈80% faster response times, ≈30% lower deployment cost via AWS SageMaker).
- Helped develop the
- April 2022 – June 2023: SDE Intern at Webknot Technologies Pvt. Ltd.
- Revamped API endpoints within the Palette project, achieving a notable 30% reduction in response times.
- Fine-tuned data flow for the DeckGL plugin within Sisense by elevating the efficiency of JAQL queries, ensuring a smoother and more responsive user experience.
- Engineered a custom plugin for Sisense BI software, enabling the seamless display of geojson data on a GeoJSON layer atop maps rendered via DeckGL.
Teaching & Mentoring
- Spring 2026: Co-mentor, ASU SCience and ENgineering Experience (SCENE)
- Co-mentored a SCENE high school student on event detection in multivariate time-series data; supported experimental design and preparation for AZSEF.
- Spring 2025: Graduate Teaching Assistant, CSE 510 Database Management Systems Implementation, ASU
- Fall 2024: Graduate Teaching Assistant, CSE 515 Multimedia and Web Databases, ASU
Awards
- University Graduate Fellowship 2026, School of Computing and Augmented Intelligence, Arizona State University
- Best Demo Award, CIKM 2024 For Introducing CausalBench: A Flexible Benchmark Framework for Causal Analysis and Machine Learning