<?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 Patrícia Oliveira on Medium]]></title>
        <description><![CDATA[Stories by Patrícia Oliveira on Medium]]></description>
        <link>https://medium.com/@psoliv?source=rss-cb60a21590fd------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*rVw_mySfC0J0W5SRJCiKJQ.jpeg</url>
            <title>Stories by Patrícia Oliveira on Medium</title>
            <link>https://medium.com/@psoliv?source=rss-cb60a21590fd------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 14:31:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@psoliv/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[ChatGPT for Software Testers: my experiment]]></title>
            <link>https://medium.com/@psoliv/chatgpt-for-software-testers-my-experiment-4090a6c43de2?source=rss-cb60a21590fd------2</link>
            <guid isPermaLink="false">https://medium.com/p/4090a6c43de2</guid>
            <category><![CDATA[automation-testing]]></category>
            <category><![CDATA[experiment]]></category>
            <category><![CDATA[software-testing]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[scripting]]></category>
            <dc:creator><![CDATA[Patrícia Oliveira]]></dc:creator>
            <pubDate>Wed, 17 May 2023 22:19:12 GMT</pubDate>
            <atom:updated>2023-05-18T10:57:57.545Z</atom:updated>
            <content:encoded><![CDATA[<p>As someone with a scientific background, I decided to do my own experiment with <a href="https://chat.openai.com/">ChatGPT</a>.</p><blockquote><strong>The Hypothesis: </strong>ChatGPT can help me become a more efficient and effective Software Tester</blockquote><blockquote><strong>The Method: </strong>Don’t read any article. Use only ChatGPT and my brain 🧠</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eaZJzW0-FYwM86GNvkRXNg.png" /></figure><h3>First Prompt</h3><blockquote>Me: “How can you help Software Testers excel at their role?”</blockquote><p>Why not use the tool itself to understand where to start? This got me plenty of topics to delve into!</p><p>I chose a few and used a similar approach to begin each one’s exploration:</p><blockquote>Me: “What’s the best way to ask you to [do something]?”</blockquote><p>This helped me understand how to refine the prompts to get to the best possible outcome, tailored to my needs.</p><p>Here goes what I learned!</p><h3>Generate Test Cases</h3><p>After asking ChatGPT for the best way to prompt it to generate test cases, these are my top 3 recommendations:</p><ol><li>Clearly specify the functionality (this one seems obvious, right? but the more specific you are and the more context you provide, the more time you’ll save in the 3rd point — trust me)</li><li>Ask for specific types of tests (functional, positive/negative, corner cases)</li><li>Review and Refine (in the absence of details it will make assumptions that may not be right! In that case, you should re-prompt it for the corrections)</li></ol><p>Now, following the tips and an actual example of a prompt (provided by the very own ChatGPT), this was what I wrote:</p><blockquote><em>Me: “Please generate </em><strong><em>10</em></strong><em> </em><strong><em>functional</em></strong><em> test cases for a </em><strong><em>document upload</em></strong><em> feature on a </em><strong><em>mobile application</em></strong><em>. Consider both </em><strong><em>valid</em></strong><em> and </em><strong><em>invalid</em></strong><em> inputs, including edge cases such as unsupported documents and corrupted files. Focus on testing the </em><strong><em>navigation</em></strong><em> and </em><strong><em>error handling</em></strong><em>. Please provide the test steps, expected results, and any preconditions or setup needed for each test case.”</em></blockquote><p>The titles of the Test Cases suggested by ChatGPT:</p><ol><li>Valid document upload</li><li>Empty document upload</li><li>Invalid document format</li><li>Large document upload</li><li>Upload cancellation</li><li>Network disruption during upload</li><li>Duplicate document upload</li><li>Document preview display</li><li>Error handling — server unavailable</li><li>Error handling — file corruption detection</li></ol><p>And the structure of one of them:</p><blockquote>Test Case 3: Invalid Document Format</blockquote><blockquote>Preconditions: User is logged in to the mobile application.</blockquote><blockquote>Test Steps:<br>Navigate to the document upload screen.<br>Select a document file in an unsupported format (e.g., TXT, XLS).<br>Tap the “Upload” button.</blockquote><blockquote>Expected Result: An error message is displayed indicating that the selected document format is not supported.</blockquote><p>Pretty good for the information I provided, right?</p><h3>Test Data Generation</h3><p>In this case my top 3 recommendations are different:</p><ol><li>Be specific about structure, format and required fields</li><li>Include variations such as ranges and constraints</li><li>Specify the quantity of entries needed</li></ol><blockquote>Me: “Given the following <strong>JSON</strong>, generate another for a <strong>Portuguese</strong> person that is <strong>more than 18 years old</strong>.</blockquote><blockquote>{<br> “first_name”: “Jane”,<br> “last_name”: “Doe”,<br> “dob”: “1990–01–31”,<br> “address”: {<br> “building_number”: “100”,<br> “street”: “Main Street”,<br> “town”: “London”,<br> “postcode”: “SW4 6EH”,<br> “country”: “GBR”<br> }<br>}’<br>“</blockquote><p>The JSON returned:</p><pre>{<br> &quot;first_name&quot;: &quot;João&quot;,<br> &quot;last_name&quot;: &quot;Silva&quot;,<br> &quot;dob&quot;: &quot;2002–05–16&quot;,<br> &quot;address&quot;: {<br>   &quot;building_number&quot;: &quot;100&quot;,<br>   &quot;street&quot;: &quot;Rua Principal&quot;,<br>   &quot;town&quot;: &quot;Lisboa&quot;,<br>   &quot;postcode&quot;: &quot;1000–001&quot;,<br>   &quot;country&quot;: &quot;PRT&quot;<br> }<br>}</pre><p>Perfect! So, if we need test data that represents many people from different locations, ChatGPT can quickly and accurately generate it for us!</p><h3><strong>Scripting</strong></h3><p>Again, another set of top 3 recommendations:</p><ol><li>Clearly state the purpose and target system</li><li>Specify the input and expected output/desired outcome</li><li>Include any preferred tools or frameworks</li></ol><p>Thinking about the Document Upload feature, I thought that it would be handy to have a script to generate test files with different sizes.</p><blockquote>Me: “Please write a <strong>bash</strong> script to run on <strong>macOS</strong> that generates <strong>5</strong> sample <strong>PDF</strong> files with different sizes (<strong>from 100kb to 10mb</strong>)”</blockquote><p>It not only generated the script but it also provided run instructions and details about the PDF files (white background, black letters, etc). I was really impressed! But the ultimate test was to see if it actually worked.</p><p>It didn’t work. First attempt to run the script, it throws an error. But do you know the best part? You can go back and ask for a fix.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/728/0*V2kbQWMPUnwslJDs" /></figure><p>Second attempt, it generated more than the 5 files initially requested. No problem, ask it for a fix again.</p><p>It worked on the third iteration. So, lesson learned: <strong>don’t rely blindly</strong> on what you get.</p><h3>Test Automation</h3><p>In the realm of test automation, I wanted to see if ChatGPT could understand public API documentation and generate automated tests accordingly. The recommended principles for this prompt align with the scripting prompt.</p><blockquote>Me: “Please write an <strong>automated</strong> test using <strong>Ruby</strong> and <strong>RSpec</strong> to cover the <strong>User Group Creation </strong>as per the following API documentation: <a href="https://api.slack.com/methods/usergroups.create">https://api.slack.com/methods/usergroups.create</a></blockquote><blockquote>Make sure to also include the verification of the <strong>response schema</strong>”</blockquote><p><strong>WOW</strong>! Now I was really impressed. The endpoint was right. The request body had the mandatory fields and two of the optional fields as well. Assertions were included for the status code and the schema, as requested. It assumed the <strong>response schema was defined on a separate file</strong>, ‘<em>user_group_create_response_schema.json’, </em>and<em> </em>gave me instructions to create the file. I asked ChatGPT to create it for me. At last, only thing I needed to do was to replace the API Token: everything else was ready.</p><p>What about testing one of the expected errors? I asked it to also cover the error ‘<em>name_already_exists’</em>. I realised <strong>the test assumed I already had at least one user group created,</strong> whose name I should replace in the test. So I asked it to create the existing user group as a setup step. This is what was added:</p><pre>  before(:all) do<br>    # Prepare request payload to create the existing user group<br>    payload = {<br>      name: &#39;existing_user_group&#39;,<br>      channels: &#39;C12345,C67890&#39;,<br>      validate: true,<br>      token: token<br>    }<br><br>    # Send API request to create the user group<br>    HTTParty.post(&quot;#{base_url}#{endpoint}&quot;, body: payload)<br>  end</pre><p>Now, keeping the context, I tried to go a step further.</p><blockquote>Me: “Please write a K6 load test for the previous endpoint”</blockquote><p>Although I did not provide any details, ChatGPT generated a load test script with specific stage definitions and performance thresholds. Those can be customised afterwards.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-dI9hZauXp1GzLYn" /></figure><p>Not bad! Another task completed with success.</p><h3>Continuous Testing and CI/CD</h3><p>Already knowing ChatGPT impressive skills on generating, explaining, troubleshooting and fixing code, I decided to do something different here.</p><p>Part of the valuable work as Test Engineers is to make processes better. It can be really time consuming and requires great knowledge and vision. Could I chat for a while with ChatGPT and get some actual improvement suggestion I could implement?</p><blockquote><strong>Me</strong>: Tell me questions I can ask you to improve my Continuous Testing and CI/CD solutions<br><strong>ChatGPT</strong>: (…) What are the recommended approaches for handling flaky tests in a CI/CD pipeline? (…)<br><strong>Me</strong>: What are the recommended approaches for handling flaky tests in a CI/CD pipeline?<br><strong>ChatGPT</strong>: (…) Retry Mechanism: Implement a retry mechanism for flaky tests (…)<br><strong>Me</strong>; How can I rerun flaky tests automatically in a gitlab pipeline?<br><strong>ChatGPT</strong>: (…)<br> test:<br> script:<br> — run_flaky_tests.sh<br> retry: 3<br>(…)</blockquote><p>ChatGPT answers were shortened, but you get the point! See how you can go from really generic to really specific in a few seconds. You can use this approach for pretty much anything that you would like to have improvement suggestions on. You can also start more specific, of course, if you already have an idea.</p><h3>Test Documentation and Reporting</h3><p>ChatGPT told me it is great at generating documentation and providing nice summaries from a lot of content (such as test results). I decided to believe in it this time. Only thing I asked was to generate a Decision Log and, once again, I was impressed. How many times have you felt that you know very well what you want to say but spend a lot of time just constructing the sentences properly? Here’s your assistance!</p><h3>Conclusions</h3><p>Hypothesis confirmed to be true! But the magic comes from your prompts, interactions and reviews! Use it wisely! 🪄</p><p><em>ℹ️ Keep in mind that </em><a href="https://chat.openai.com/"><em>ChatGPT</em></a><em> training data only goes up until September 2021.<br>ℹ️ Be careful of the information you are sharing with the tool: do not include any personal and sensitive data.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4090a6c43de2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Biomedical Engineers in Software Quality Assurance]]></title>
            <link>https://medium.com/ubiwhere/the-added-value-that-biomedical-engineers-bring-to-software-quality-assurance-16a9b8464ca2?source=rss-cb60a21590fd------2</link>
            <guid isPermaLink="false">https://medium.com/p/16a9b8464ca2</guid>
            <category><![CDATA[skills]]></category>
            <category><![CDATA[quality-assurance]]></category>
            <category><![CDATA[software-testing]]></category>
            <category><![CDATA[biomedical-engineering]]></category>
            <dc:creator><![CDATA[Patrícia Oliveira]]></dc:creator>
            <pubDate>Mon, 17 Feb 2020 17:24:41 GMT</pubDate>
            <atom:updated>2020-02-19T16:09:02.438Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why they are a perfect match</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tdksOGodh81Yi78L1dA0Ug.jpeg" /></figure><p><em>I’m a Biomedical Engineer and I’ve been working as a Quality Assurance Engineer for the past 6 years. I knew very little about the role when I started. However, I was able to learn and adapt very quickly to it, often producing out-of-the-box results by having a different (but still engineering-related) mindset for quality.</em></p><p>This article will dig into the reasons why I think the Biomedical Engineering background was the main responsible for that!</p><p><strong>Intersection between different areas/interests</strong></p><ul><li>Biomedical Engineering extends across different knowledge areas. Often professionals have a hard time communicating with others, specialized in different areas, because they are not speaking the same language. That’s where biomedical engineers have a crucial role! It is also an essential skill of QAs to <strong>avoid miscommunication</strong> between different parts and help ensure the<strong> design, development, delivery and maintenance</strong> of a<strong> solution that fits the problem</strong>.</li></ul><p><strong>Pursuit for better results</strong></p><ul><li>Understanding biological processes is a hard task. Analysing them from an engineering point of view is even harder! It is required to go the extra mile to discover how the process can be changed and what different results can be achieved. That analytic skill is exactly what is expected from a QA regarding software <strong>processes and their improvement.</strong></li></ul><p><strong>Understanding the stakeholders needs</strong></p><ul><li>Biomedical Engineers have <strong>special sensibility to understand people’s needs and how they can be addressed by technology</strong>. However, despite being thrilled with the technical solution they won’t lose focus on the problem. In fact, that will be their priority.</li></ul><p><strong>Good habits from the lab</strong></p><ul><li>They are used to investigate, follow instructions and report the conclusions from strict (and expensive) lab experiments. You can expect them to be <strong>curious, organized and meticulous </strong>in their work.</li></ul><p><strong>Learning and being flexible</strong></p><ul><li>Their knowledge is distributed across distinct areas. The disadvantage of not being specialists, turns out to be a great advantage when they have to learn more about a topic or decide to specialize. Their background makes them eager and <strong>quick learners</strong> and, after all, <strong>flexible professionals</strong>. <em>Isn’t that a must have skill for the Tech area?</em></li></ul><p>I’m very proud to see Biomedical Engineers revealing to be great quality assurance professionals (and not only!). Throughout my career I’ve already seen them shine on so many different roles (software development, business development, regulatory affairs, project management, algorithms, product support, design, machine learning, research…). It is a very special profile, isn’t it?</p><h3><em>After all, the motivation is </em><strong><em>technology improving people’s lives…</em></strong></h3><p>…and we are privileged to be able to choose how to contribute. I chose to be QA and I’m currently ensuring the quality of the <strong>Smart Cities</strong> solutions we deliver at <a href="https://www.ubiwhere.com/"><strong>Ubiwhere</strong></a><strong>.</strong> Have you thought about how we could use the knowledge of human biological systems to improve the efficiency and management of our cities? Well, <em>that’s a good topic for another article</em>. :)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=16a9b8464ca2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ubiwhere/the-added-value-that-biomedical-engineers-bring-to-software-quality-assurance-16a9b8464ca2">Biomedical Engineers in Software Quality Assurance</a> was originally published in <a href="https://medium.com/ubiwhere">Ubiwhere</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>