<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Yenuli Munasinghe on Medium]]></title>
        <description><![CDATA[Stories by Yenuli Munasinghe on Medium]]></description>
        <link>https://medium.com/@yenulimunasinghe04?source=rss-90e5add17a4f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*TQyLr5CSRiqY2nCw</url>
            <title>Stories by Yenuli Munasinghe on Medium</title>
            <link>https://medium.com/@yenulimunasinghe04?source=rss-90e5add17a4f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 13:13:33 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@yenulimunasinghe04/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Day 12 — Making FinFlow’s Financial Data More Reliable]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-12-making-finflows-financial-data-more-reliable-5e2a86bb7624?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/5e2a86bb7624</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Tue, 03 Nov 2026 10:11:25 GMT</pubDate>
            <atom:updated>2026-11-03T10:11:25.332Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ryRxsHax8N7j92kBS81B1w.png" /><figcaption>AI- generated image</figcaption></figure><h4>After working on the backend foundation for the AI Financial Insight Generator, I continued improving the financial data flow in FinFlow today.</h4><p>While working on the previous implementation, I realized that generating useful financial insights depends heavily on how accurately the application handles budgets and transactions. So before moving further with the AI feature, I wanted to make sure the underlying financial logic was reliable.</p><p>One of the main things I worked on was event budget creation. I updated the backend to allow users to create an event together with its budget items. I also added validations to make sure the allocated amounts don’t exceed the overall event budget and that the selected categories actually exist.</p><p>I also spent time improving transaction validation and the approval workflow. Expenses now need to be associated with the correct budget items, and transactions that exceed their allocated budgets require a written justification. I also improved how revision requests and resubmissions are handled.</p><p>Another important part of today’s work was testing these changes with more realistic financial data.</p><p>I created a scenario with an event budget of Rs. 450,000, seven budget categories, and twelve transactions. I manually calculated the expected results and compared them with the application’s calculations.</p><p>The result was encouraging. All seven database tests passed, and every financial calculation matched my expected values. Along with these, all 69 unit tests also passed.</p><p>I also encountered a couple of interesting issues along the way. Some database operations were failing because of short transaction timeouts, and when Gemini was unavailable, insight requests were taking around 60 seconds to respond.</p><p>After investigating, I adjusted the database transaction time limits and introduced controlled retry handling for Gemini failures. This reduced the response time in the failure scenario to around three seconds.</p><p>It reminded me that integrating an external AI service also means thinking about what happens when that service doesn’t work as expected.</p><h3>What’s Next?</h3><p>The backend improvements are coming together, but there is still more to do.</p><p>My next focus is building the frontend screens for managing event budget items and connecting the existing financial insights functionality to the dashboard.</p><p>I’m also waiting to verify the live Gemini integration once the model is available again.</p><p><strong>One thing I learned today , </strong>Before making an application intelligent it’s important to make sure the data and business logic behind it are reliable.</p><p>More improvements coming to FinFlow!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5e2a86bb7624" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 11 — Building the Backend for FinFlow’s AI-Powered Financial Insight Generator]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-11-building-the-backend-for-finflows-ai-powered-financial-insight-generator-18b1d387f6d3?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/18b1d387f6d3</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Sat, 24 Oct 2026 14:32:43 GMT</pubDate>
            <atom:updated>2026-10-24T14:32:43.222Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*glfVo-52XjGD3qaFiLecsQ.png" /><figcaption>AI generated image</figcaption></figure><p>In my previous article, I shared my plan to introduce an AI-powered Financial Insight Generator to FinFlow.</p><p>The idea came from noticing that even though the system provides financial reports and dashboards, event organizers and committee members still need to manually analyze the information to understand how an event performed financially.</p><p>Today, I started turning that idea into an actual feature by working on its backend.</p><p>My main focus was to build a foundation that could retrieve real financial data, perform the necessary calculations, and generate useful financial insights.</p><h3>Starting with Financial Calculations</h3><p>Before integrating AI, I wanted to make sure the system could calculate and organize the financial information correctly.</p><p>Since financial data involves actual monetary values, I didn’t want to depend on an AI model to perform the calculations.</p><p>I started by implementing a financial calculation engine that analyzes event budgets, income, and expenses.</p><p>It currently calculates,</p><ul><li>Approved and pending income and expenses.</li><li>Remaining and projected budgets.</li><li>Budget utilization.</li><li>Net cash flow.</li><li>Spending breakdowns by category.</li></ul><p>I also introduced different budget states, such as safe, warning, near-limit, at-limit, and overspent, to help identify the financial condition of an event.</p><p>This was an important part of the implementation because the insights generated later need to be based on accurate financial values.</p><h3>Generating Insights Without AI</h3><p>Another thing I worked on was generating financial insights using predefined rules.</p><p>For example, if an event exceeds its planned budget, the system can identify the overspending and generate a warning. Similarly, if pending expenses could cause the event to exceed its budget, the system can highlight that risk.</p><p>I also added category-specific recommendations and general budget health messages.</p><p>This approach allows the application to provide useful financial information even without depending entirely on an external AI service.</p><h3>Connecting Real Financial Data</h3><p>After working on the calculation logic, I connected the new module to FinFlow’s existing database using Prisma.</p><p>The system can now retrieve event details, budget items, expense categories, and transactions from Supabase.</p><p>I also handled situations such as repeated category entries, missing events, and database failures.</p><p>Getting this part working was important because I wanted the feature to analyze actual application data rather than just hardcoded sample values.</p><p>I also confirmed that the Supabase database connection was working correctly after resuming the database.</p><h3>Integrating Gemini AI</h3><p>The next step was introducing the AI component.</p><p>For this, I used the <strong>Google Gen AI SDK with Gemini 3.8 Flash</strong>.</p><p>The idea is to provide the model with the calculated financial summary and let it generate understandable explanations, important findings, and recommendations.</p><p>I also used <strong>Zod</strong> to validate the structured responses returned by the model.</p><p>One thing I considered during implementation was what would happen if the AI service failed or returned an invalid response.</p><p>To handle this, I included a fallback mechanism that uses the rule-based insights. This means the feature can still return useful information even when Gemini is unavailable.</p><p>Another important consideration was data privacy. Instead of sending raw transaction records or user details to the AI model, the integration uses prepared financial summaries.</p><p>This helped me understand that integrating AI into an application involves more than simply making an API request.</p><h3>API Integration and Testing</h3><p>I created a protected API endpoint for retrieving event financial insights.</p><p>The endpoint requires JWT authentication and connects the financial calculations, database queries, and AI integration.</p><p>After completing the backend implementation, I ran several checks to verify that everything was working as expected.</p><p>It was good to see the different parts working together after completing the initial implementation.</p><h3>What’s Next?</h3><p>Although the backend foundation is now implemented, the feature is not fully completed yet.</p><p>I still need to work on persistent event budget-item management, the real financial data entry flow, and the frontend dashboard for displaying the generated insights.</p><p>My next focus will be connecting this backend functionality to the user interface so event organizers and committee members can actually view and use the financial insights within FinFlow.</p><h3>What I Learned Today</h3><p>One of the main things I realized while working on this feature is that AI should not be responsible for everything.</p><p>For financial applications, accurate calculations and reliable data handling are especially important. The AI model can help explain the information and generate recommendations, but the actual financial values should come from the application’s own calculation logic.</p><p>I also got hands-on experience working with an AI API, structured responses, response validation, and fallback handling.</p><p>There is still more to implement, but today’s progress gave me a better understanding of how an AI-powered feature can be integrated into an existing full-stack application.</p><p><strong>Day 11 takeaway:</strong> Building an AI-powered feature is not just about connecting to an AI model. It is also about preparing reliable data, handling failures, and making sure the application remains useful even when the AI service is unavailable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18b1d387f6d3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 10 — Planning an AI-Powered Financial Insight Generator for FinFlow]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-10-planning-an-ai-powered-financial-insight-generator-for-finflow-1f7074a4e723?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f7074a4e723</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Tue, 20 Oct 2026 13:21:32 GMT</pubDate>
            <atom:updated>2026-10-20T13:21:32.496Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aSFWRs5QPVxOcsH38PW6fA.png" /><figcaption>AI generated image</figcaption></figure><p>After completing the main workflow of FinFlow, I started thinking about what else could be improved to make the system more useful for its users.</p><p>One thing I noticed was that even though the system provides financial reports, budget comparisons and dashboards, event organizers and committee members still need to manually go through the information to understand the financial performance of an event.</p><p>For example, after an event, they may need to check whether the expenses stayed within the planned budget, which categories consumed the most money, where overspending happened and what could be improved when organizing future events.</p><p>Although FinFlow already helps organize and present this information, analyzing the results and identifying useful insights still requires manual effort.</p><p>This made me think about adding another feature to the system.</p><h3>The Idea</h3><p>I decided to explore adding an <strong>AI-powered Financial Insight Generator</strong> to FinFlow.</p><p>The main idea is to help event organizers and committee members understand their financial data without having to manually analyze every detail.</p><p>Instead of only displaying numbers and charts, I want the system to generate a simple explanation of the financial performance of an event.</p><p>For example, if an event exceeded its planned budget, the system could identify the categories that contributed to the additional expenses and provide suggestions that could help with future budget planning.</p><h3>What I’m Planning to Build</h3><p>The feature will mainly focus on analyzing event-related financial information.</p><p>Some of the things I want it to do include:</p><ul><li>Comparing planned budgets with actual income and expenses.</li><li>Identifying categories where spending exceeded the planned amount.</li><li>Highlighting important spending patterns.</li><li>Generating a simple financial summary after an event.</li><li>Providing recommendations that could help with planning future events.</li></ul><p>The goal is not to replace the existing financial reports and dashboards, but to make the information they provide easier to understand and use.</p><h3>My Planned Approach</h3><p>Since FinFlow already stores information about events, budgets and transactions, I plan to use that existing data for the new feature.</p><p>My initial idea is to retrieve the relevant financial information through the backend, calculate the necessary financial values and then integrate an AI model through an API to generate insights based on those results.</p><p>The generated insights will then be displayed in the FinFlow dashboard.</p><p>I also need to consider how to make sure the AI uses accurate financial information and does not generate misleading conclusions.</p><p>For now, I’m planning to start with individual event analysis before considering more advanced features.</p><h3>What I Want to Learn</h3><p>This will be my first attempt at integrating an AI-powered feature into one of my existing full-stack applications.</p><p>Through this, I want to understand how AI APIs can be integrated with a backend, how to provide the necessary data to an AI model and how to handle the generated responses within an application.</p><p>I also want to learn about the challenges that come with using AI, especially when working with financial information where accuracy is important.</p><h3>What’s Next</h3><p>My next step is to review FinFlow’s existing database structure and financial reporting logic to identify what information will be needed for the insight generator.</p><p>After that, I’ll start working on the backend integration and gradually connect the feature to the dashboard.</p><p>There may be challenges along the way, but I think this will be a useful opportunity to explore how AI can be used to improve an existing application rather than building something completely new.</p><p><strong>Day 10 takeaway:</strong> Presenting financial information is useful, but helping users understand that information and make better decisions can make the system even more valuable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f7074a4e723" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 9 — Bringing FinFlow Together]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-9-bringing-finflow-together-1ebc6b448374?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ebc6b448374</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Sun, 20 Sep 2026 15:16:01 GMT</pubDate>
            <atom:updated>2026-09-20T15:16:01.373Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R8lZEVIE35NB_CCnihdhfg.png" /></figure><p>Today was the ninth day of working on <strong>FinFlow</strong>.</p><p>After working on the different parts of the system separately, today I focused on bringing everything together and testing the complete workflow.</p><h3>What I Worked On</h3><p>I tested the full financial process starting from <strong>creating an event and setting its budget</strong>.</p><p>A committee member can then submit an expense with the required details and receipt.</p><p>The system checks the effect of the expense on the relevant budget and identifies when it goes over the planned amount.</p><p>The transaction can then be reviewed by the President, who can approve or reject it.</p><p>Once it is approved, the financial information is reflected in the relevant reports and dashboard.</p><p>I also checked that the action is recorded in the <strong>audit log</strong> so that the transaction history can be tracked.</p><h3>What I Learned</h3><p>Testing the complete workflow helped me understand the project much better than testing each feature separately.</p><p>When all the parts are connected, a small change in one part of the system can affect another part.</p><p>For example, approving a transaction is not just a status change. It can also affect the budget, financial totals, reports and audit records.</p><h3>Problems I Faced</h3><p>One of the main things I had to check was whether all the different parts of the workflow were working together correctly.</p><p>I had already tested individual features, but testing the complete process helped me find areas that needed further checking and improvement.</p><h3>How I Solved It</h3><p>I went through the workflow step by step and checked the result after each stage.</p><p>I tested the transaction submission, budget checking, approval, financial calculations and audit logging to make sure the information was being carried through the system correctly.</p><p>I also completed the deployment setup so that FinFlow could be accessed through the deployed application rather than only from my local environment.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 9, I had a much clearer view of the complete FinFlow workflow.</p><p>I started with the project setup, then moved through database design, user roles, transactions, budgets, approvals, reports, authentication and finally deployment.</p><p>There are still things I want to improve in the future, but completing the main workflow has given me a much better understanding of the project as a whole.</p><p>Some future ideas I have include supporting multiple societies, adding bank statement reconciliation and generating PDF invoices.</p><p><strong>Day 9 takeaway: Testing the complete workflow helped me understand how all the individual parts of a system work together.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ebc6b448374" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 8 — Hardening Authentication and Route Protection]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-8-hardening-authentication-and-route-protection-0fec2704a51b?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/0fec2704a51b</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 16:01:01 GMT</pubDate>
            <atom:updated>2026-09-19T16:01:01.556Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*kAWbuROby1aJBFsbqQgu6Q.jpeg" /><figcaption>AI generated image</figcaption></figure><p>Today was the eighth day of working on <strong>FinFlow</strong>.</p><p>After moving the application towards the cloud, I focused on making sure that users could not access parts of the system without being properly authenticated and authorized.</p><h3>What I Worked On</h3><p>I started by removing some of the temporary login and demo shortcuts that I had used while developing the application.</p><p>Instead of starting with a default user or using temporary login checks, users now need to go through the actual authentication process.</p><p>I also worked on the <strong>route protection</strong> in the frontend.</p><p>Pages such as the dashboard, approvals and reports should not be accessible to someone who has not logged in.</p><p>I added authentication checks so that unauthenticated users are redirected away from these protected pages.</p><h3>Improving Login</h3><p>I also updated the login process so that users can log in using either their <strong>Member ID or email address</strong>.</p><p>The backend checks the credentials against the database instead of using the temporary values that were used during earlier development.</p><h3>Role-Based Approval</h3><p>Another important part I worked on was making sure that only the <strong>President</strong> can approve transactions.</p><p>The backend now checks the user’s role before allowing the approval endpoint to be accessed.</p><p>This is important because hiding an approval button from a user is not enough. The backend should also prevent the action if someone tries to access the endpoint directly.</p><h3>What I Learned</h3><p>Today I learned more about the difference between <strong>authentication and authorization</strong>.</p><p>Authentication answers the question of who the user is, while authorization decides what that user is allowed to do.</p><p>I also understood that security checks should be enforced on the backend and not only on the frontend.</p><h3>Problems I Faced</h3><p>One thing I had to be careful about was removing the temporary development shortcuts without breaking the normal login and navigation flow.</p><p>I also had to make sure that users with different roles could still access the parts of the system that they were supposed to use.</p><h3>How I Solved It</h3><p>I connected the login process with the actual database users and used the existing <strong>JWT and RolesGuard</strong> setup to protect the backend endpoints.</p><p>I also added frontend route protection so that users cannot simply navigate directly to a protected page without logging in.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 8, the authentication flow was much closer to how I wanted it to work in the actual application.</p><p>The temporary shortcuts were removed, protected routes were added and the President-only approval rule was enforced on the backend.</p><p>Tomorrow, I’ll bring everything together and test the complete financial workflow.</p><p><strong>Day 8 takeaway: Security should be enforced by the system, not just assumed from the user interface.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0fec2704a51b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 7 — Moving FinFlow to the Cloud]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-7-moving-finflow-to-the-cloud-d52bd5cb1a67?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/d52bd5cb1a67</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Fri, 18 Sep 2026 15:00:24 GMT</pubDate>
            <atom:updated>2026-09-18T15:00:24.582Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tGgYdkHncqf9TCHOsA3lTg.png" /><figcaption>AI generated image</figcaption></figure><p>Today was the seventh day of working on <strong>FinFlow</strong>.</p><p>After testing the system across different devices, I started looking at how I could move the project from my local development environment to the cloud.</p><h3>What I Worked On</h3><p>I worked on moving the database to <strong>Supabase PostgreSQL</strong>.</p><p>Until now, the database setup had mainly been part of my local development environment. Moving it to the cloud would allow the deployed application to access the database without depending on my own computer.</p><p>I also updated the <strong>Prisma configuration</strong> to work with PostgreSQL.</p><p>One thing I learned during this process was that there are different ways to connect to a cloud database.</p><p>I configured separate connection strings for the normal database connection and the direct connection used for database operations such as migrations.</p><h3>Setting Up the Backend</h3><p>I also started working on deploying the <strong>NestJS backend on Vercel</strong>.</p><p>Since Vercel uses a serverless approach, the NestJS application needed some changes to work correctly in that environment.</p><p>I created the required serverless setup and configured the routing so that requests could reach the backend API.</p><h3>Problems I Faced</h3><p>The cloud setup was not as straightforward as I expected.</p><p>I had to deal with differences between the local database setup and the cloud PostgreSQL environment.</p><p>I also encountered an npm dependency issue while preparing the backend for deployment.</p><h3>How I Solved It</h3><p>I configured Prisma with the appropriate PostgreSQL connection settings and separated the pooled database connection from the direct connection.</p><p>For the dependency issue, I used the appropriate npm configuration so that the required packages could be installed without blocking the deployment setup.</p><h3>What I Learned</h3><p>Today I learned that moving an application to the cloud is not simply about uploading the project.</p><p>The database connections, environment variables, server configuration and deployment environment all need to work together.</p><p>I also learned why <strong>connection pooling</strong> is important when working with serverless applications, since many short-lived serverless functions can otherwise create too many database connections.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 7, I had started moving FinFlow from a local development setup towards a <strong>cloud-based application</strong>.</p><p>The database was connected to Supabase and the backend was being prepared for deployment on Vercel.</p><p>Tomorrow, I’ll focus more on <strong>authentication and protecting the application routes</strong>.</p><p><strong>Day 7 takeaway: Moving a project to the cloud involves more than just deploying the code.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d52bd5cb1a67" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 6 — Taking FinFlow Beyond Localhost]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-6-taking-finflow-beyond-localhost-095320adb4dc?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/095320adb4dc</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Wed, 16 Sep 2026 17:32:30 GMT</pubDate>
            <atom:updated>2026-09-16T17:32:30.214Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o47iNflB4-VNlgGdLgkAZQ.jpeg" /></figure><p>Today was the sixth day of working on <strong>FinFlow</strong>.</p><p>Until now, I had mainly been testing the system on my own computer. Today, I wanted to see how FinFlow would work when accessed from another device, especially a mobile phone connected to the same network.</p><h3>What I Worked On</h3><p>I started working on making the backend accessible through the <strong>local network</strong> instead of only through localhost.</p><p>By default, the backend was only accessible from the same computer. This meant that even if my phone was connected to the same Wi-Fi network, it couldn’t communicate with the backend.</p><p>To solve this, I changed the NestJS server to listen on <strong>0.0.0.0</strong>, which allows it to accept connections from other devices on the network.</p><p>I also updated the <strong>CORS configuration</strong> so that requests from the local network could communicate with the backend.</p><h3>Testing on Mobile</h3><p>After making these changes, I started testing FinFlow on my phone.</p><p>This also made me notice some UI problems that were not as obvious when using the system on a larger screen.</p><p>I worked on making the navigation more suitable for smaller screens by adding a <strong>collapsible mobile sidebar</strong>.</p><p>I also adjusted some of the transaction details and voucher preview areas so that they would work better on mobile devices.</p><h3>What I Learned</h3><p>Today I learned that a system can look completely fine on a desktop but feel very different when used on a mobile device.</p><p>Testing on a real phone helped me notice things such as screen space, navigation and how information is displayed in smaller areas.</p><p>It also helped me understand why testing on different devices is important instead of only testing on my own computer.</p><h3>Problems I Faced</h3><p>The main problem was that the mobile device could not access the backend even though both devices were connected to the same Wi-Fi network.</p><p>The issue was not with the frontend itself, but with how the backend server was listening for connections.</p><h3>How I Solved It</h3><p>I changed the NestJS server to listen on <strong>0.0.0.0</strong> and updated the CORS settings to allow the local network origin.</p><p>After that, I was able to access the application from my phone and continue testing the responsive UI.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 6, FinFlow was no longer limited to being tested only on my computer.</p><p>I could access it from different devices on the same network and also had a better idea of what needed to be improved for mobile users.</p><p>Tomorrow, I’ll start working on <strong>moving FinFlow towards the cloud</strong>.</p><p><strong>Day 6 takeaway: Testing on real devices can reveal problems that are easy to miss on a desktop.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=095320adb4dc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 5 — Turning Financial Data into Useful Information]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-5-turning-financial-data-into-useful-information-d7a41deccfec?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/d7a41deccfec</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 02:59:46 GMT</pubDate>
            <atom:updated>2026-09-15T02:59:46.402Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g60zCmiV9i5TK50BVGj5NA.jpeg" /><figcaption>AI generated image</figcaption></figure><p>Today was the fifth day of working on <strong>FinFlow</strong>.</p><p>After working on the database, transactions, budgets and approval workflow, today I started focusing on how the financial information can be presented in a more useful way.</p><h3>What I Worked On</h3><p>I worked on the <strong>financial reports and dashboard</strong> part of FinFlow.</p><p>One of the main things I wanted to show was the difference between what was planned and what actually happened.</p><p>For example, an event can have an expected income and estimated expenses before it happens. After the event, the system can compare these values with the actual income and expenses.</p><p>I also worked on things such as <strong>budget utilization, remaining budget and event profitability</strong>.</p><p>This makes it easier to understand the financial result of an event instead of only looking at individual transactions.</p><h3>What I Learned</h3><p>Today I learned that storing financial data is only one part of building a useful system.</p><p>The data should also be presented in a way that helps users understand the current financial situation.</p><p>For example, knowing that an event spent Rs. 80,000 is useful, but comparing that with the planned budget gives much more information.</p><h3>Some Changes I Made</h3><p>I also started making FinFlow more general instead of keeping it limited only to university societies.</p><p>The idea is that the system could also be useful for different <strong>societies, clubs and organizations</strong> that need to manage their finances.</p><p>I also faced a port conflict while running the project on Windows. I changed the frontend port and updated the CORS configuration so that the frontend and backend could communicate correctly.</p><h3>Problems I Faced</h3><p>One thing I had to be careful about was making sure the reports use the correct financial information.</p><p>For example, pending or rejected transactions should not be treated as finalized financial records.</p><h3>How I Solved It</h3><p>I made the financial calculations based on the appropriate approved records.</p><p>This allows the reports and dashboard to give a more accurate view of the financial situation.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 5, FinFlow had moved quite a bit beyond the initial project setup.</p><p>I started with the project foundation, then worked on the database, user permissions, transactions, approvals and finally the financial reports.</p><p>There is still a lot more to improve, but I now have a much clearer idea of how the different parts of FinFlow will work together.</p><p><strong>Day 5 takeaway: Financial data becomes more useful when it can help us understand what is happening.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d7a41deccfec" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 4 — Working on Approvals and Audit Logging]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-4-working-on-approvals-and-audit-logging-19a3cc04238c?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/19a3cc04238c</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 17:19:20 GMT</pubDate>
            <atom:updated>2026-09-14T17:19:20.851Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HopDyajcocX-152CvpBjhg.jpeg" /><figcaption>ai generated image</figcaption></figure><p>Today was the fourth day of working on <strong>FinFlow</strong>.</p><p>After working on transaction submission yesterday, today I focused on what should happen <strong>after a transaction is submitted</strong>.</p><h3>What I Worked On</h3><p>I started working on the <strong>transaction approval workflow</strong>.</p><p>The idea is that submitted transactions should be reviewed before they become part of the official financial records.</p><p>I worked on an approval area where an authorized user can view the transaction details, receipt and its effect on the relevant budget.</p><p>I also worked on showing the budget impact when a transaction goes over the planned amount.</p><p>Another part I worked on was the <strong>audit log</strong>.</p><p>The system records important actions such as transaction submissions, approvals, rejections and other changes so that there is a record of what happened.</p><h3>What I Learned</h3><p>Today I learned that a financial system needs more than just the ability to add and update transactions.</p><p>There should also be a proper process for <strong>reviewing and approving</strong> financial information.</p><p>I also understood why an audit log is useful. If something changes in the system, it should be possible to know what happened and who made the change.</p><h3>Problems I Faced</h3><p>One thing I had to think carefully about was how to handle transactions that go over the planned budget.</p><p>Automatically blocking every over-budget transaction would not always make sense because there could be valid reasons for the additional expense.</p><h3>How I Solved It</h3><p>Instead of automatically rejecting these transactions, I added an <strong>over-budget indication</strong> and a place to provide a justification when the additional spending is approved.</p><p>I also connected the approval actions with the audit log so that important decisions can be tracked.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 4, FinFlow had a clearer process for <strong>submitting, reviewing and approving transactions</strong>.</p><p>This helped me understand that financial management is not only about recording numbers. There also needs to be a proper process behind those numbers.</p><p>Tomorrow, I’ll focus on <strong>financial reports and dashboards</strong>.</p><p><strong>Day 4 takeaway: Financial information needs to be reviewed and tracked, not just recorded.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=19a3cc04238c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 3 — Connecting Events, Budgets and Transactions]]></title>
            <link>https://medium.com/@yenulimunasinghe04/day-3-connecting-events-budgets-and-transactions-39331f7b5be5?source=rss-90e5add17a4f------2</link>
            <guid isPermaLink="false">https://medium.com/p/39331f7b5be5</guid>
            <dc:creator><![CDATA[Yenuli Munasinghe]]></dc:creator>
            <pubDate>Sat, 12 Sep 2026 14:28:49 GMT</pubDate>
            <atom:updated>2026-09-12T14:28:49.961Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZKnkIXAebr-jQueFYfkGGQ.jpeg" /><figcaption>AI generated image</figcaption></figure><h3>Today was the third day of working on <strong>FinFlow</strong>.</h3><p>After working on the database structure and user roles, I started moving towards the actual financial workflow of the system.</p><h3>What I Worked On</h3><p>I started working on the <strong>event planning and budgeting</strong> part of FinFlow.</p><p>The idea is to allow users to create an event and plan the expected expenses for different areas such as Food, Transport, Printing and other event-related costs.</p><p>I also worked on the transaction submission flow for both <strong>income and expenses</strong>.</p><p>The transactions can be connected with the relevant event and budget information so that the system can later compare the actual spending with the planned amount.</p><p>I also worked on <strong>receipt uploads</strong> using Cloudinary so that supporting documents can be attached to financial transactions.</p><h3>What I Learned</h3><p>Today I learned that budgeting and transaction management need to work together.</p><p>If an event has a planned budget, the system should not only record how much was spent, but also be able to compare that amount with what was originally planned.</p><p>This helped me understand how the different parts of FinFlow are starting to connect with each other.</p><h3>Problems I Faced</h3><p>One thing I had to think about was how to identify when a transaction goes beyond the planned budget.</p><p>If the system only records the expense amount, it would not be possible to properly understand its effect on the event budget.</p><h3>How I Solved It</h3><p>I connected the transaction information with the relevant budget and event details.</p><p>This allows the system to calculate the budget impact and identify when an expense goes over the planned amount.</p><p>I also added some fallback data during development so that I could continue testing the features while working on the database connection.</p><h3>What I’m Taking Forward</h3><p>By the end of Day 3, I had a clearer connection between <strong>events, budgets and transactions</strong>.</p><p>The system is starting to move from just storing information towards actually following a financial workflow.</p><p>Tomorrow, I’ll continue working on the <strong>approval process and audit logging</strong>.</p><p><strong>Day 3 takeaway: The budget should be connected to the actual spending, not treated as a separate part of the system.</strong></p><p><strong>Day 3 takeaway: The budget should be connected to the actual spending, not treated as a separate part of the system.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=39331f7b5be5" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>