Inspiration

Financial crime teams review thousands of alerts but still miss coordinated fraud rings. I wanted to shift investigations from transactions → networks → action.

What it does

Tasvi identifies coordinated account behavior, assigns explainable risk scores, and surfaces a priority investigation queue so analysts know who to act on first—and why.

How we built it

Multi-source transaction data is unified, enriched with graph-based risk intelligence, exported as Hyper files, and investigated through Tableau Cloud dashboards.

Challenges we ran into

Balancing realistic data volume with clear, human-friendly investigation workflows—showing meaningful risk without overwhelming the investigator.

Accomplishments that we're proud of

Built a pipeline-first system, used Hyper API over CSVs, and turned Tableau into an action-oriented investigation command center.

What we learned

Effective fraud detection depends on relationships, not rules—and trust comes from clear explanations, not black-box scores.

What's next for Tasvi (Financial Crime Investigation Center)

Near-real-time ingestion, automated alerts, deeper network visualization, and enterprise-scale investigation workflows.

Built With

  • networkx
  • pandas
  • python
  • tableau-cloud
  • tableau-desktop
  • tableau-hyper-api
Share this project:

Updates