01 · Challenge
Citizens can report visible river pollution, but many reporting systems provide little transparency about verification, responsibility, or whether the issue was ever resolved.
Back to work
AI First Hackathon · River accountability
Nadi Rakshak
Nadi Rakshak helps citizens document visible river pollution, applies transparent AI confidence scoring, maps reports publicly, and tracks each case from Reported to Resolved.
Scroll into system
Case study brief
The problem, then the product response.
02 · Response
I designed Nadi Rakshak as an evidence-led civic product that links geo-tagged reports, transparent AI confidence, human review, a public map, and accountable case states.
System anatomy
How the system coordinates work.
The architecture is expressed as responsibilities and boundaries so the model, workflow logic, interfaces, and human controls remain inspectable.
01
Citizen reporting flow for geo-tagged evidence and structured incident details
02
Transparent confidence layer with human review and false-positive safeguards
03
Public heatmap and status workflow from Reported through Verified and Resolved
Implementation
What I built and the decisions behind it.
01
Mapped the reporting journey from evidence capture through verification and resolution.
02
Specified confidence explanations and false-positive safeguards instead of presenting AI output as a verdict.
03
Designed a public heatmap, escalation path, responsible-party workflow, and sensor/data roadmap.
04
Documented risks, entities, state transitions, and an implementation scaffold.
Evidence & validation
What an evaluator can inspect.
Product narrative, pitch material, risk matrix, and scoring specification.
Data schemas, accountability states, interface evidence, and project media.
Hackathon presentation and professional project publication.
Reflection
For civic AI, public trust depends less on a prediction score than on visible evidence, review rights, status ownership, and a traceable path to resolution.
Next iteration
- →Pilot with a small verified dataset and local stakeholders.
- →Add moderation and appeal workflows.
- →Measure report quality, verification time, and closure rate.
Outcomes
What the project delivered.
Pitch, product narrative, risk matrix, and confidence-scoring specification
Sensor roadmap, data schemas, accountability states, and implementation scaffold
Public visibility and traceable resolution designed into the core workflow
Technology & concepts
01AI Vision02Geospatial UX03Civic Tech04Product Research05Responsible AI