FinDS 2026 | Data Management for Modern Financial Systems

Topics of Interest

Topics of interest include but are not limited to the following...

• Financial knowledge modeling: Modeling the evolving financial data to trace money laundering, model systemic risk, and perform causal analysis.
• (Near) Real-time Fraud Detection: Problems such as real-time subgraph matching in high-velocity streams. Eg:- capturing sophisticated fraud-rings.
• Graph-RAG: Addressing the core data management challenges of moving beyond simple vector retrieval to enable multi-hop logical reasoning over financial knowledge graphs.

• Foundational Storage: Efficient storage while ensuring durability, security & encryption, fast access, and the use of distributed file systems, cloud-based data lakes, and scalable database architectures.
• Optimized Processing: Parallel processing, In-memory computation, optimized query engines to handle workloads and keep latency in check. Real-time Anomaly detection for streaming data.

• Agentic AI Infrastructure: Scalable Management systems for millions of autonomous payment agents.
• Algorithmic Governance: Designing data-centric systems for creating the "black box recorder" for financial AI, essential for dispute resolution, regulatory audits etc.
• Dynamic Knowledge Graphs: Representing agent capabilities and beliefs in real-time, enable complex (near) real-time queries over evolving financial systems.

• Hardware Co-design: The co-design of data systems with FPGAs, Trusted Execution Environments (TEEs), and modern hardware to accelerate cryptographic workloads, privacy-preserving computations, and graph-based fraud detection etc.
• Faster stream-processing: System architectures designed for complex feature engineering and model inference on massive transaction streams under extreme latency constraints.
• Native Support: Architectures supporting real-time, multi-hop graph traversal and complex analytics within high-throughput transactional workloads.

• Privacy-Preserving ML: Support for training and deploying models on sensitive financial data, including techniques such as federated learning and efficient analytics over encrypted data.
• Data Management for Federated Learning (FL): Infrastructure to manage the entire FL lifecycle, addressing challenges such as secure and efficient data partitioning, ensuring statistical consistency, and optimizing communication costs across different legal and geographical jurisdictions.

Full Research Papers (8 pages): For Mature, well developed, and rigorously evaluated research
Short/Vision Papers (4 pages): For early-stage results, novel system designs, or bold, provocative ideas.

Template: Kindly adhere to the submission template guidlines found here.
You can find the accepted Overleaf format here.
Author names: You can use the following LaTeX command to compile your paper without author names:

\documentclass[sigconf, anonymous, review]{acmart}
Submissions that do not follow these requirements will be desk-rejected.

Evaluation: Reviewing will be double-anonymous, for which the submissions must be anonymized by following the same anonymity requirements as for regular track papers at the SIGMOD/PODS 2026 conference.

To provide maximum visibility and foster deep discussions, every accepted paper (both full and short) will receive presentation opportunities: An oral presentation slot within a technical session. A spot in a dedicated interactive poster session facilitating direct engagement and in-depth conversations between authors and attendees.
Accepted papers will be non-archival and posted on the workshop website.