Open to engineering roles and client projects
Work / Venture / Tafity Receipt
Case study · Venture

Tafity Receipt.

The core receipt-processing product on the main Tafity platform, turning receipt photos into structured business data.

A product under Tafity, the main platform at tafity.com.

RoleSoftware Engineer; receipt processing, AI providers, API and billing
ForBusinesses and developers using the Tafity platform
TimelineMay – Aug 2026
StackPHP, MySQL, Gemini, M-Pesa, Paystack, Stripe
StatusLive · Visit ↗
Tafity Receipt Tafity Receipt on a phone

Public product website

400+receipts processed
3AI provider integrations
The problem

A receipt photo is useful evidence but difficult to search, report on or connect to another system. Manually entering every item adds work, particularly when receipts arrive in batches.

What I built

Receipt data people and software can use

I built receipt upload and background processing, integrated swappable AI providers, and exposed the extracted data through search, Excel exports and a public REST API. API keys and credit packages support developer access, with M-Pesa, Paystack and Stripe billing.

  • ExtractionReceipt details and individual line items become structured data.
  • Bulk processingUpload receipts in batches and process them in the background.
  • Export & APISearch data, export it to Excel or retrieve it using an API key.
  • BillingCredit packages with M-Pesa, Paystack and Stripe payments.
Receipt processing workflow on the public website
Receipt processing workflow on the public website
Architecture

A processing pipeline with interchangeable AI providers

Background workers handle extraction outside the web request. The provider can be switched between Gemini, Google Vision and OpenAI. Extracted receipt and line-item data can be retrieved through the platform or its API.

Tafity Receipt workflow
  1. Receipt uploadPhotos enter the Tafity platform.
  2. AI extractionBackground processing turns receipt details into structured data.
  3. Export & APIThe results can be searched, exported or retrieved through the API.
Key decisions

Choices that made it hold up.

  • Swappable AI providers

    Gemini, Google Vision and OpenAI can be selected without tying receipt processing to one provider.Supports provider changes as cost and extraction quality vary.

  • Background processing

    Workers handle extraction outside the web request.Keeps longer processing work out of the interactive request.

Evidence

Production usage

CheckWhat it doesResult
ReceiptsPhotos processed into structured data400+ receipts processed
Batch uploadMultiple receipts processed togetherBulk upload supported
Export and integrationExcel output and API-key accessAvailable on Tafity
Contact

An engineering role or a project to build?

Share the opportunity or the problem you need solved. I usually reply within a day with questions, an approach or a time to talk.

Phone+254745600377
Based inNairobi, Kenya · working worldwide
What do you need?
You can also mention a budget in your preferred currency below.