Abhinav Gorantla

Abhinav Gorantla

Ph.D. student in Computer Science, ASU

Abhinav Gorantla

I am a first-year 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), and published work on causally-informed data retrieval at SIGMOD 2026. I’m currently exploring how causal priors can improve the training and generalization of ML models. I also build the systems behind this research — I’m a core developer of the CausalBench platform and co-presented a tutorial on causal benchmarking at KDD 2025.

Research

My research lies at the intersection of causal machine learning, multi-objective optimization, and generative models. I am especially interested in:

  • 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
    • Advisor: Prof. K. Selçuk Candan
  • Master of Science in Computer Science, Arizona State University, Dec 2025
    • GPA: 4.00/4.00
    • Advisor: Prof. K. Selçuk Candan
  • Bachelor of Technology in Computer Science and Engineering, Vellore Institute of Technology, May 2023
    • GPA: 8.98/10.00

Experience

  • August 2024 - Now: Graduate Research Assistant at Emitlab, School of Computing and Augmented Intelligence, ASU
    • Developing an optimized algorithm for efficient Skyline retrieval in relational database systems.
    • Collaborating with researchers at CASCADE Lab to maintain and improve causalbench.org, a platform dedicated to causal discovery benchmarks.
    • Built a causal-analysis recommendation system; led end-to-end integration and serverless deployment on AWS Lambda; published at CIKM 2025.
  • August 2024 - May 2025: Graduate Teaching Assistant at School of Computing and Augmented Intelligence, ASU
    • Assisted in CSE515 and CSE510 graduate-level computer science courses.
  • March 2024 – August 2024: Graduate Services Assistant at Emitlab, School of Computing and Augmented Intelligence, ASU
    • Helped develop the causalbench Python 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).
  • 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.

Publications

  1. P. Mandal, A. Gorantla, K. S. Candan, and M. L. Sapino.
    “Causal Search for Skylines (CSS): Causally-Informed Selective Data De-Correlation.”
    Accepted to the ACM SIGMOD/PODS International Conference on Management of Data (SIGMOD 2026).
    [preprint]
  2. A. Kapkıç, P. Mandal, A. Gorantla, S. Wan, E. Çoban, P. Sheth, H. Liu, and K. S. Candan.
    “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, Nov 10–14, 2025, pp. 6426–6431.
    [paper]
  3. A. Kapkıç, P. Mandal, A. Gorantla, S. Wan, E. Çoban, P. Sheth, H. Liu, and K. S. Candan.
    “CausalBench: Causal Learning Research Streamlined.”
    Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’25), Vol. 2,
    Toronto, Canada, Aug 3–7, 2025, pp. 6239–6240.
    [paper] [tutorial site] [slides] [video]
  4. A. Kapkıç, P. Mandal, S. Wan, P. Sheth, A. Gorantla, Y. Choi, H. Liu, and K. S. Candan.
    “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, 2024. (🏆 Best Demo Paper Award)
    [paper] [website]

Tutorials

  • CausalBench: Causal Learning Research Streamlined – Tutorial at KDD 2025, Toronto, Canada.
    Overview of causal machine learning, challenges in benchmarking, and hands-on use of the CausalBench platform.
    [tutorial site] [slides] [video]

Community Service & Outreach

  • Student Volunteer, ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025)

Projects

  1. Research Publications Analysis tool
    • Proposed an architecture and built a research publications analysis tool for ASU. This tool was built as a web application which could fetch research paper information affiliated with ASU using SCOPUS APIs and perform a text analysis on their abstracts.
    • Reduced the server response time by 80% and improved the user experience by integrating RabbitMQ message queues in the system.
    • Tech stack used: ReactJS, NodeJS, Python-FastAPI, RabbitMQ, MongoDB, AWS S3, AWS Sagemaker, OpenAI APIs.
  2. Multimodal Image Retrieval System using Advanced Feature Analysis and Search Techniques
    • Developed a Python-based image retrieval engine encompassing feature extraction from Caltech101 dataset images, latent semantics computation, clustering, and classification.
    • Employed Locality Sensitive Hashing to index image features, optimizing nearest neighbor searches and ensuring scalability for expansive image datasets.
  3. Enhancing Diversity in the LLM Modulo Framework through Multi-Response Generation
    • Developed the Diversified LLM Modulo framework to address looping and redundancy in the LLM Modulo framework.
    • Improved the performance of the LLM Modulo Framework on Planning tasks. Tested my framework on the Google Deepmind Natural Plan benchmark and achieved a performance improvement of 300% by increasing the diversity of LLM (Large Language Model) Responses.