While working on a client project that required accessing their APIs through Cloudflare Access, we encountered a critical issue. After requesting IP whitelisting for our office WiFi IP address, the client successfully whitelisted it. However, once we started implementing the APIs into the application, we discovered that even with internet access, the API calls themselves were not happening when we hit the login API.

Through thorough research and investigation, I identified that a CA/SSL certificate needed to be installed on the device accessing the API. This was a crucial step that was initially overlooked. By installing the appropriate Cloudflare Origin CA certificate on the development device, we were able to establish secure connections and successfully make API calls through Cloudflare Access.

Our application required offline functionality for handling attachments. The challenge arose because the attachment content we were receiving was in Base64 format. Storing such large Base64 data directly into the SQLite database was causing significant performance issues and database bloat, making the application slow and inefficient.

I developed an optimized approach where, before inserting data into the SQLite database, we convert the Base64 image content into file system URLs using the device's file system. This approach significantly improved performance and reduced database size. When retrieving the data for display, we reverse the process by converting the file system URLs back to Base64 format, ensuring seamless functionality while maintaining optimal performance.

Users needed to enter various documents into the application such as passports, driving licenses, certificates, safety permits, and other identification documents. The client required a verification system to ensure that the entered documents are authentic and belong to the user who is submitting them. They needed basic verification to check if the entered driving license or passport actually relates to the user, and only allow access if verified.

I built an MVP that integrates AI technology with a custom-built tool into the application. The solution uses a POST API that processes captured document images. When a user captures an image, it's converted to Base64 format. Our custom AI tool then extracts the user's name from the document and cross-verifies it with the user's entered card details in the application. Additionally, the system validates expiration dates for passports and driving licenses, providing alerts if documents are expired or about to expire (with a threshold of below 60 days). By using our own custom tool instead of paid OCR services, we reduced costs while maintaining accuracy. This MVP provided the client with proper usage and use cases, allowing them to optimize and refine the solution based on real-world feedback.

The application required extracting receipt details from various types of receipts. Initially, using traditional OCR services would have incurred significant costs for processing each receipt. The client needed a cost-effective solution that could extract all necessary information from receipts and streamline the expense generation process.

I developed a cost-effective solution by integrating an API with built-in AI capabilities using OpenAI. Instead of paying for expensive OCR services, I created a POST API that processes receipt images. When users capture or upload a receipt from their mobile device, the content is sent to the API which extracts all relevant information using AI. The extracted data is then automatically populated into the expense form fields, eliminating manual data entry. After processing, the system automatically sends an email notification to the customer confirming that the receipt has been processed. This solution provides a single-click expense generation feature, significantly reducing the time and effort required for expense management while keeping costs low.