Sajjadur Rahman
Adobe Founders Tower, San Jose, CA 95113

I am a Senior Applied Science Manager at Adobe, where I lead the Evaluation and Continuous Learning charter within Adobe Experience Platform (AEP). My work focuses on building large-scale evaluation frameworks for enterprise-grade agentic systems and enabling continuous learning through scaling supervision and adaptive learning strategies within the Adobe Agent Orchestrator. Previously, I led the Center for Excellence for AI Quality within AEP, helping enable the general availability of Adobe Agent Orchestrator and Agents and leading the quality program for the GA launch of Adobe Brand Concierge.

My work synthesizes techniques from data management, AI, and HCI to design scalable, interactive, and reliable systems that power enterprise AI assistants, enable continual learning in agentic systems, and support AI-assisted collaborative workflows. Prior to Adobe, I was the Founding Research Manager of the Data-AI Symbiosis group at Megagon Labs. I received my PhD from the University of Illinois at Urbana–Champaign, where I worked with Aditya Parameswaran. My work has been published in premier conferences in Databases (SIGMOD and VLDB), HCI (CHI and CSCW), and NLP (EMNLP and NAACL), recognized with awards (best demo award at ICDE 2018), featured in popular technology blogs, and deployed in open-source as well as enterpris systems.

News

Selected Projects

AI Quality for Enterprise AI Assistant

Quality for Enterprise

We introduce a scalable and principled methodology for benchmark management, evaluation, error analysis, and continual improvement strategy for an enterprise AI assistant for customer experience orchestration. By adopting this holistic framework, organizations can systematically enhance the reliability and performance of their AI Assistants.

Agentic System for Enterprise Data

Agentic for Enterprise

We outline our approaches toward understanding and implementing a more effective agentic workflow in the wild. To achieve the goal, we draw on the cognitive science concepts of System 1 (fast, intuitive thinking) and System 2 (slow, deliberate, analytical thinking.) We instantiate the vision in an open-source plattform for authoring enterprise-grade agentic workflows: Blue.

Human-AI Collaborative Design

Human-AI collaboration for creative design

We built MixTAPE, a mixed-initiative system, for human-AI collaboration for creative tasks with checklists and action plans as the basis for coordination. The tool enabled AI-driven task management for designers, web developers, and customer success managers. Deployed within B12, as of June 2021 (two years since it's launch), MixTAPE has helped create more than 60k todos across more than 2.5k projects.

Selected Publications

Synergistic Activities

2026    - DASHSys@VLDB'26 (General Chair), ACL'26 (Industry PC)
2025    - SIGMOD'26 (PC), VLDB Journal (reviewer)
2024    VLDB'24 (PC), SIGMOD'25 (PC), Reviewer:NLP4HR@EACL'24, TiiS
2023    MATCHING@ACL 2023 (Program Chair), MATHCING@ACL 2023 (Panel Moderator), ACL 2023 (PC), EMNLP 2023 (PC), BLP@EMNLP 2023 (PC), SIGMOD 2023 (Demo PC), BigVIS 2023 (PC), CHI 2023 (Reviewer)
2022    EMNLP 2022 (PC), DaSH@EMNLP 2022 (PC), ISSRE 2022 (Industry PC), BigVIS 2022 (PC), WIT@ACL (PC), CHI 2022 (Reviewer)
2021    SIGMOD 2022 (PC), VLDB 2021 (Panelist), BigVIS 2021 (PC), WIT@KDD (PC), CHI 2021 (Reviewer)
2020    SIGMOD 2021 (PC)
2019    VLDB 2019 (Demo, external reviewer)
2018    IIT 2018 (PC)

Papers by Research Themes

Agentic AI for Enterprise

  • Building, Serving, and Growing a Conversational AI Assistant for Enterprise SIGMOD Industry Track, 2026.
  • MageSQL: Enhancing in-context learning for Text-to-SQL applications with large language models AIDB Workshop, VLDB 2025.
  • Effectiveness of Prompt Optimization in NL2SQL Systems NOVAS@SIGMOD 2025.
  • Retrieval Augmented Generation in the Wild: A System 2 Perspective IEEE Data Engineering Bulletin, 2025.
  • Demonstration of a multi-agent framework for text to sql applications with large language models CIKM 2024.
  • A Blueprint Architecture of Compound AI Systems for Enterprise Compound AI Systems Workshop at Data+AI Summit 2024.
  • Reasoning Capacity in Multi-Agent Systems: Limitations, Challenges and Human-Centered Solutions Compound AI Systems Workshop at Data+AI Summit 2024.

    • Evaluation and Continual Learning

  • Evaluation and incident prevention in an enterprise AI assistant. AI Magazine, 2025.
  • CYPHERBENCH: Towards Precise Retrieval over Full-scale Modern Knowledge Graphs in the LLM Era. ACL, 2025.
  • From Facts to Folklore: Evaluating Large Language Models on Bengali Cultural Knowledge AACL-IJCNLP.
  • Factlens: Benchmarking fine-grained fact verification Findings of ACL 2025.
  • [Workshop] CMDBench: A Benchmark for Coarse-to-fine Multimodal Data Discovery in Compound AI Systems. Guide-AI@SIGMOD 2024.
  • Characterizing LLMs as Rationalizers of Knowledge-intensive Tasks. Findings of ACL 2024.
  • Low-resource Interactive Active Labeling for Fine-tuning Language Models. Findings of EMNLP 2022.
  • Characterizing practices, limitations, and opportunities related to text information extraction workflows: a human-in-the-loop perspective. CHI 2022.
  • Low-resource Entity Set Expansion: A Comprehensive Study on User-Generated Text Findings of NAACL 2022.
  • Benchmarking Spreadsheet Systems. SIGMOD 2020.
  • [Workshop] Understanding Data Analysis Workflows on Spreadsheets: Roadblocks and Opportunities. HILDA@SIGMOD 2020.

    • AI-assisted Collaborative Work

  • Lakevisage: Towards scalable, flexible and interactive visualization recommendation for data discovery over data lakes VLDB 2025.
  • MEGAnno+: A Human-LLM Collaborative Annotation System EACL 2024.
  • Human-LLM collaborative annotation through effective verification of LLM labels. CHI 2024.
  • Towards Transparent, Reusable, and Customizable Data Science in Computational Notebooks CHI 2023 LBW.
  • [Demo] Weedle: Composable Dashboard for Data-centric NLP in Computational Notebooks WWW 2023.
  • [Vision]Toward multifaceted human-centered AI HCAI@NeurIPS 2022.
  • NOAH: Interactive Spreadsheet Exploration with Dynamic Hierarchical Overviews. VLDB 2021.
  • [Workshop] Towards integrated, interactive, and extensible text data analytics with LEAM. DASH-LA@NAACL 2021.
  • Leam: An Interactive System for In-situ Visual Text Analysis. CIDR 2021.
  • MixTAPE: Mixed-initiative Team Action Plan Creation Through Semi-structured Notes, Automatic Task Generation, and Task Classification. CSCW 2020.
  • [Demo] Faster, Higher, Stronger: Redesigning Spreadsheets for Scale. ICDE 2019.
  • I’ve Seen “Enough”: Incrementally Improving Visualizations to Support Rapid Decision Making. VLDB 2017.
  • SeeDB: Efficient data-driven visualization recommendations to support visual analytics. VLDB 2015.