What Fleet Operators Were Dealing With
Fleet operators across North America face the same grind. Vehicles pass through multiple networks, and each network bills separately. Every network has its own file format, its own field names, its own date conventions. Some send CSV. Others send fixed-width text. A few send XML with no schema documentation.
Without automation, staff download files manually from each network's portal, reformat them in spreadsheets, and try to match charges to rental agreements by vehicle and date. The process is slow and error-prone. Charges end up attributed to the wrong customer, or not attributed at all. Vehicles that are not registered with a network get billed at the higher unregistered rate. Citations arrive weeks after the original event, by which point the rental has closed and the original driver is nearly impossible to trace.
There is no audit trail. When a customer disputes a charge, nobody can trace it back to the original source file with any confidence. The kludge of spreadsheets and manual cross-referencing does not scale.
A SaaS company saw the opportunity to solve this for fleet operators across the market. They needed an engineering partner to design and build the platform from scratch.
A Platform Built Around the Rental Agreement
Sequoia Applied Technologies is a Santa Clara software engineering firm that builds data platforms and cloud systems for SaaS companies, logistics firms, and enterprise software businesses. This engagement was a greenfield build: design and deliver a platform that could ingest charge files from any network, normalise them into a consistent format, match each charge to the correct rental, and produce invoices that fleet operators could trust.
The critical design decision was treating the rental agreement as the central entity. Every other data domain converges on the agreement: the fleet roster, the vehicle registrations, the charge transactions, the billing statements, the citations. That single anchor point is what makes reliable attribution possible at scale. A charge is not just linked to a vehicle. It is linked to a vehicle during a specific rental window, which means it is linked to a specific customer and a specific contract.
The platform accepts charge files however each network delivers them. Authority-specific ingestion functions parse each format and translate it into a unified internal schema. Malformed records are caught during validation and flagged rather than silently corrupting downstream data. Clean records flow into a normalised layer, then into an enriched layer where they are matched against the rental agreement database by vehicle and transaction time. Unmatched charges are surfaced for review rather than discarded.
Zero manual reconciliation steps in the end-to-end workflow. Files arrive, get processed, matched, and billed without anyone touching a spreadsheet.
Vehicle registration is automated too. When a vehicle enters or leaves a fleet, the platform registers or deregisters it with every relevant network. This keeps the entire fleet on transponder rates rather than the higher unregistered cash rates.
Serverless, Event-Driven, and Built to Scale
The platform runs entirely on Azure with no permanent compute infrastructure to manage. Azure Functions handle all processing, spinning up on demand when files arrive or events fire, and scaling back to zero when idle. Azure Service Bus provides the messaging backbone, with dedicated queues for each stage of the pipeline so a surge in file arrivals does not create backpressure on billing or fleet processing.
Node.js functions handle file ingestion, normalisation, agreement matching, registration sync, and billing generation. Each function has a single responsibility and communicates through the message bus. The architecture is event-driven: a file landing in storage triggers processing without polling or scheduled jobs.
Queues isolate each pipeline stage. Messages are retried up to 10 times before routing to a dead-letter queue for investigation. A 10-minute deduplication window prevents double-processing during network retries. Topic and subscription patterns allow a single event to fan out to multiple downstream processes.
PostgreSQL stores fleet rosters, rental agreements, charge transactions, and billing records. The schema follows a medallion architecture: raw data lands first, normalised data sits in a second layer, enriched and matched data in a third. Every record carries provenance metadata back to its source file.
No credentials are stored in application code or environment variables. All Azure-to-Azure communication uses Managed Identity token exchange. External secrets like SFTP keys for network file exchange live exclusively in Azure Key Vault. The platform cannot leak what it does not hold.
Infrastructure is defined as code using Bicep templates. A single parametrised template generates a fully correct, fully isolated environment for development, QA, or production from one deployment command. Environments are structurally identical. Only names and scales differ. This eliminates the drift that had plagued earlier attempts at automation.
What the Platform Delivers to Fleet Operators
Fleet operators using the platform see immediate gains in time savings and billing accuracy. Staff who previously spent hours downloading files and reformatting spreadsheets can focus on work that actually requires human judgment. Billing disputes drop because every charge traces back to a specific source file through a verifiable chain.
Two outcomes stand out as particularly consequential. The first is citation resolution velocity. Without automation, a violation notice arriving weeks after a rental closed often goes unmatched. It either escalates into a larger penalty or gets absorbed as a loss. With automatic matching against the rental history, operations teams can respond to citations in hours rather than days, disputing legitimate errors and processing valid charges before penalty thresholds are crossed.
The second is the flexibility of the billing model. Different rental brands within the same corporate group often carry different billing policies. The platform handles this through configuration rather than code. Convenience fees, exchange rates for international operations, corporate account exemptions: all configurable per brand. When an operator adds a new brand or restructures its policy, the change takes an afternoon instead of a development cycle.
100% of fleet vehicles maintained on transponder rates across all networks. No more cash-rate surcharges from unregistered plates.
The full audit trail from source file to customer invoice changes how finance teams work. Disputed charges can be investigated in minutes rather than hours. The data provenance is unambiguous: here is the file, here is the record, here is when it was processed, here is the agreement it matched to, here is the invoice line. No more guesswork.
Common Questions About Fleet Billing Platforms
What does a fleet billing reconciliation platform actually do?
It automates the process of collecting charge files from multiple networks, normalising them into a consistent format, matching each charge to the correct rental agreement by vehicle and date, and producing customer invoices with a full audit trail. The goal is to eliminate the manual gruntwork of downloading files from different portals, reformatting them in spreadsheets, and cross-referencing rental records by hand.
How do you handle files that arrive in different formats from each network?
Each network sends data in its own structure, with different field names, date formats, and encodings. The platform includes authority-specific ingestion functions that parse each format and translate it into a unified internal schema. Malformed records are caught during validation and reported back to the source rather than silently corrupting downstream data.
What technology stack powers this kind of platform?
The platform Sequoia built runs on Azure. Azure Functions written in Node.js handle compute, Azure Service Bus provides the messaging backbone for event-driven processing, and PostgreSQL stores the data. Security uses Managed Identity throughout, meaning no credentials are stored in application code. Infrastructure is defined as code using Bicep templates, so environments are reproducible and drift-free.
How does the platform keep vehicle registrations current across all networks?
When a vehicle enters or leaves the fleet, the platform automatically registers or deregisters it with every relevant network. This ensures vehicles are always billed at the registered transponder rate rather than the higher unregistered cash rate. The registration logic is triggered by fleet roster changes, not by manual tracking.
Can the platform handle multiple brands with different billing policies?
Yes. The billing model is configurable per brand. Convenience fees, exchange rates for international operations, and corporate account exemptions are all handled through configuration rather than code changes. When a new brand is added or a policy changes, the adjustment is made to the billing rules without touching the processing pipeline.
What kind of companies does Sequoia Applied Technologies work with on fleet and logistics software?
Sequoia Applied Technologies is a Santa Clara, California software engineering firm that works with fleet operators, logistics companies, and enterprise software firms. Engagements include new platform builds, data pipeline engineering, cloud architecture on Azure and AWS, and production delivery for high-volume transactional systems.