01 · Manual toll and violation evidence
Business teams combined DIMTS records, NPCI files, spreadsheets, and toll-plaza evidence manually. Violation files and clarifications moved between teams and plazas over email.
Customer work · Banking and mobility
A FASTag data platform connected toll transactions, vehicle-class detection, and violation workflows
BluePi connected toll transactions, NPCI files, vehicle images, automated processing, reporting, and violation workflows in one AWS-based platform. The system reduced manual reporting and supported vehicle-class validation and reconciliation.
FASTag data, vehicle evidence, and violation work entered one platform
FASTag growth increased transaction volume, file processing, reporting, settlement, and violation work. BluePi connected the data lake, operational reporting, vehicle-class model, concessionaire portal, and case-management workflow.
02 · Data and violation platform
An AWS data platform ingested transaction changes and NPCI files, prepared refined datasets, supported querying and dashboards, automated violation processing, and connected vehicle-class detection to a maker-checker workflow.
03 · Reviewed settlement exceptions
Manual reporting and validation work decreased, vehicle-class evidence became part of the violation process, and settlement and revenue-leakage exceptions became easier to identify.
From toll transaction to reviewed violation
Transaction and NPCI data entered the AWS data platform, operational reporting exposed exceptions, vehicle-class evidence supported violation review, and the maker-checker workflow carried each case through clarification and resolution.
Unify transaction and file ingestion
DIMTS changes and NPCI files entered the same data platform while retaining ingestion and refined storage layers.
Automate the vehicle-class check
The model compared camera evidence with tagged and mapped vehicle classes so clear cases could be pre-classified before business review.
Move exceptions through a case workflow
Maker, checker, and toll-plaza users worked from one case record instead of exchanging violation files and clarifications over email.
Where this pattern fits
This pattern fits mobility and financial workflows where a physical event, a transaction record, external files, and an adjustment workflow must agree. It also applies to parking, access control, logistics checkpoints, and asset movement.
Case details
Open a section to review the customer problem, implementation, business change, and architecture.
01The starting point
The organization needed to manage growing FASTag volumes while replacing manual reporting, violation, and settlement work.
- Separate data sources: DIMTS transactions, NPCI files, local spreadsheets, and toll-plaza evidence were difficult to combine.
- Manual processes: Reporting, violation handling, settlement, and file validation depended on Excel and Access.
- Limited history: Business teams could not easily query or analyze historical files across systems.
- Vehicle-class disputes: Camera evidence and FASTag vehicle classes required repeated manual comparison.
- Offline coordination: Business teams and toll plazas exchanged violation files and clarifications over email.
02What BluePi built
BluePi connected FASTag ingestion, processing, discovery, reporting, vehicle-class detection, and case management.
- Oracle CDC: Oracle GoldenGate and Amazon MSK captured DIMTS transaction changes.
- NPCI file ingestion: SFTP files entered through services on Amazon EKS.
- Lake storage: Amazon S3 stored ingestion and refined Apache Hudi tables.
- Processing and orchestration: Amazon EMR and Apache Airflow automated data preparation and FASTag processes.
- Discovery: Apache Atlas recorded source, structure, relationships, refresh, and lineage.
- Query and reporting: Presto and Apache Superset supported analysis and dashboards.
- Concessionaire portal: AWS Fargate, Spring Boot, and React supported reporting and plaza access.
- Violation preprocessing: PySpark and Airflow prepared violation files and images for validation.
- Vehicle-class model: Databricks and Python trained and served batch image classification.
- Case management: Flowable coordinated maker, checker, plaza, and plaza-head-office actions.
03Results
The platform reduced manual processing and connected vehicle evidence, transaction data, reporting, and violation workflows.
- Less manual reporting and file preparation
- Vehicle-class validation supported by computer vision
- Violation and settlement exceptions became visible in shared workflows
- Toll-plaza communication moved from email into a case-management portal
- Historical FASTag data became available for querying and dashboards
- Better visibility supported reduced revenue leakage and more complete toll accounting