<?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 Thuvarahan on Medium]]></title>
        <description><![CDATA[Stories by Thuvarahan on Medium]]></description>
        <link>https://medium.com/@thuvarahan2003?source=rss-83315a8036e3------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*pmivJN53xx8bNOvZbX6sLA.png</url>
            <title>Stories by Thuvarahan on Medium</title>
            <link>https://medium.com/@thuvarahan2003?source=rss-83315a8036e3------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 13:34:40 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@thuvarahan2003/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[Understanding Password Recovery in WSO2 Identity Server]]></title>
            <link>https://medium.com/@thuvarahan2003/understanding-password-recovery-in-wso2-identity-server-176ec95d937b?source=rss-83315a8036e3------2</link>
            <guid isPermaLink="false">https://medium.com/p/176ec95d937b</guid>
            <category><![CDATA[java]]></category>
            <category><![CDATA[identity-management]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[wso2]]></category>
            <dc:creator><![CDATA[Thuvarahan]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 08:40:10 GMT</pubDate>
            <atom:updated>2026-09-15T08:40:10.696Z</atom:updated>
            <content:encoded><![CDATA[<h4>A practical look at how password recovery requests, confirmation links, and notification flows work in WSO2 Identity Server</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8OSPBHAx9l66hr8Z8Cp-Vw.png" /></figure><p>Password recovery looks simple from a user&#39;s point of view.</p><p>You forget your password, click <strong>Forgot Password</strong>, receive an email or OTP, verify yourself, and create a new password.</p><p>But while exploring WSO2 Identity Server, I realized that there is much more happening behind that small flow.</p><p>A password recovery system needs to identify the correct user, create secure recovery codes, select a notification channel, generate a valid recovery link, verify the recovery request, and finally allow the password to be changed safely.</p><p>I recently spent some time exploring this flow inside WSO2 Identity Server, especially around its User Account Recovery APIs and notification handling. This article is a simple explanation of what I learned.</p><h3>Password recovery is more than sending an email</h3><p>The main purpose of password recovery is obvious: allow a legitimate user to regain access to an account when they no longer know the password.</p><p>The important word here is <strong>legitimate</strong>.</p><p>A password recovery mechanism effectively provides another path into an account. If that path is poorly designed, it can become a security weakness.</p><p>A good recovery flow therefore needs to answer several questions:</p><ul><li>Which account is being recovered?</li><li>How can the user prove that they own that account?</li><li>How long should a recovery code remain valid?</li><li>Where should recovery information be sent?</li><li>How should the reset request be verified?</li><li>What happens if somebody tries to reuse an old recovery code?</li></ul><p>WSO2 Identity Server provides self-service password recovery mechanisms through options such as email links and OTP-based verification. Administrators can configure the available recovery channels and expiry settings from the Identity Server console.</p><h3>Looking at the flow through the API</h3><p>One useful way to understand password recovery is to look at it as a sequence of API operations rather than one large action.</p><p>WSO2 Identity Server provides a User Account Recovery API v2 for this purpose. The password recovery process is divided into several stages rather than exposing one endpoint that immediately changes the password.</p><p>At a high level, the flow looks like this:</p><pre>User forgets password<br>        ↓<br>Identify the user<br>        ↓<br>Start password recovery<br>        ↓<br>Choose notification channel<br>        ↓<br>Receive recovery information<br>        ↓<br>Confirm the recovery request<br>        ↓<br>Receive reset authorization<br>        ↓<br>Set a new password</pre><p>Breaking the process into multiple stages is useful because each stage has a different security responsibility.</p><h3>Step 1: Initiating password recovery</h3><p>The first step is identifying the account that the user wants to recover.</p><p>The v2 recovery API exposes:</p><pre>POST /api/users/v2/recovery/password/init</pre><p>The request contains claims that can identify the user. Additional properties can also be supplied as metadata.</p><p>A simplified request could look like this:</p><pre>{<br>  &quot;claims&quot;: [<br>    {<br>      &quot;uri&quot;: &quot;http://wso2.org/claims/givenname&quot;,<br>      &quot;value&quot;: &quot;user1&quot;<br>    }<br>  ]<br>}</pre><p>If the user can be identified and password recovery is available, WSO2 can return information about the recovery methods that the user can use.</p><p>For example, depending on the configuration, the available channels may include:</p><pre>EMAIL<br>SMS<br>EXTERNAL</pre><p>The response can also contain recovery codes and information about the next step in the flow.</p><p>An interesting design decision here is that the API does not immediately reset anything.</p><p>At this stage, the system is only establishing:</p><blockquote><em>“This account exists, and these are the recovery methods available for it.”</em></blockquote><h3>Step 2: Requesting the recovery notification</h3><p>After selecting a recovery channel, the next part of the flow uses:</p><pre>POST /api/users/v2/recovery/password/recover</pre><p>This request includes the recovery code and the channel that should be used.</p><p>For example:</p><pre>{<br>  &quot;recoveryCode&quot;: &quot;example-recovery-code&quot;,<br>  &quot;channelId&quot;: &quot;1&quot;<br>}</pre><p>The API can then send recovery information through the selected notification channel.</p><p>For an email-based flow, this is where the user may receive the email containing the information needed to continue the recovery process.</p><p>For external channels, the API can instead return a confirmation code to the caller.</p><p>This is where I found the notification side of the system particularly interesting.</p><p>The API layer is only one part of the process. Behind it, recovery information needs to be passed through the appropriate recovery services and eventually into the notification system.</p><p>A simplified view is:</p><pre>Recovery API<br>      ↓<br>Password recovery service<br>      ↓<br>Recovery information<br>      ↓<br>Notification event<br>      ↓<br>Email / SMS / external channel</pre><h3>Notification templates are part of the actual feature</h3><p>Email notifications are sometimes treated as just presentation.</p><p>In authentication systems, they are much more important than that.</p><p>A password recovery email may contain a link that includes values generated during the recovery process.</p><p>Conceptually, a template may need information such as:</p><pre>confirmation code<br>username<br>tenant<br>service provider<br>callback information</pre><p>These values are supplied when the notification is generated and are used to build the final message seen by the user.</p><p>This means that the contract between the application logic and the notification template matters.</p><p>If a template expects a property but the backend never supplies a value for it, the resulting email can become invalid even though the recovery code itself is perfectly correct.</p><p>While exploring the WSO2 recovery implementation, this was one of the most useful lessons for me: <strong>a bug in an authentication flow does not always exist inside the authentication algorithm itself. It can exist at the boundary between two components.</strong></p><h3>Optional metadata can still affect the final result</h3><p>The account recovery API allows additional metadata to be passed using properties.</p><p>In some recovery scenarios, values such as a service provider or callback can be relevant when generating the final recovery experience.</p><p>For example, imagine an application initiating password recovery and expecting the user to return to a particular application afterward.</p><p>Conceptually:</p><pre>Application<br>     ↓<br>Password Recovery Request<br>     ↓<br>WSO2 Identity Server<br>     ↓<br>Recovery Email<br>     ↓<br>Confirmation<br>     ↓<br>Return to application</pre><p>Now consider that some of those values are optional.</p><p>A robust implementation needs to handle both situations:</p><pre>Property exists<br>→ use the supplied value</pre><p>and:</p><pre>Property does not exist<br>→ handle the absence safely</pre><p>It should never accidentally expose an internal template variable to the end user.</p><p>This sounds like a small implementation detail, but it demonstrates an important general principle:</p><blockquote><em>Optional input still requires explicit behaviour when it is absent.</em></blockquote><p>That principle applies far beyond password recovery.</p><h3>Step 3: Confirming the recovery request</h3><p>Receiving an email or OTP should not itself give permission to change the password.</p><p>The next stage verifies the confirmation information.</p><p>WSO2 exposes:</p><pre>POST /api/users/v2/recovery/password/confirm</pre><p>The confirmation code, and an OTP where applicable, is validated at this stage.</p><p>When validation succeeds, the server returns a reset code that can be used in the final password-reset operation.</p><p>So the system separates two ideas:</p><pre>&quot;I received the recovery information&quot;</pre><p>from:</p><pre>&quot;I am now authorized to reset this password&quot;</pre><p>That distinction is important.</p><h3>Step 4: Resetting the password</h3><p>The final stage uses:</p><pre>POST /api/users/v2/recovery/password/reset</pre><p>At this point, the request includes the information produced by the earlier stages together with the new password.</p><p>Conceptually:</p><pre>{<br>  &quot;resetCode&quot;: &quot;...&quot;,<br>  &quot;flowConfirmationCode&quot;: &quot;...&quot;,<br>  &quot;password&quot;: &quot;new-password&quot;<br>}</pre><p>Only after the necessary recovery state has been established and confirmed does the password actually change. The API can also reject a new password when it violates configured password policies.</p><p>The complete process therefore becomes:</p><pre>INIT<br> ↓<br>RECOVER<br> ↓<br>CONFIRM<br> ↓<br>RESET</pre><p>That sequence made the architecture much easier for me to understand.</p><h3>Why recovery codes need an expiry</h3><p>Another important part of recovery is expiration.</p><p>A password recovery link should not remain usable forever.</p><p>WSO2 allows administrators to configure the expiry duration of password recovery links or OTPs.</p><p>This limits the amount of time in which stolen or accidentally exposed recovery information can be useful.</p><p>For example:</p><pre>Recovery link generated<br>        ↓<br>Valid for configured period<br>        ↓<br>Used successfully<br>     OR<br>Expires</pre><p>The same idea appears in many security systems: temporary credentials should remain temporary.</p><h3>Callback validation matters too</h3><p>Recovery flows sometimes need to return users to another application.</p><p>But accepting arbitrary callback URLs can introduce security risks.</p><p>An identity system should not blindly trust a destination supplied by a client. Older WSO2 Identity Server documentation, for example, describes configuring a callback URL regular expression for recovery scenarios rather than allowing unrestricted callback destinations.</p><p>This highlights another useful security principle:</p><pre>Input from client<br>      ↓<br>Validate<br>      ↓<br>Use</pre><p>not:</p><pre>Input from client<br>      ↓<br>Trust immediately</pre><p>Even something as simple as a URL deserves validation when it participates in an authentication or recovery flow.</p><h3>What exploring the WSO2 codebase taught me</h3><p>Before looking into this feature, I mostly thought of password recovery as:</p><pre>Generate token → Send email → Reset password</pre><p>After following the WSO2 implementation and APIs, my mental model became much more detailed:</p><pre>Identify user<br>      ↓<br>Determine recovery options<br>      ↓<br>Generate recovery state<br>      ↓<br>Select communication channel<br>      ↓<br>Build notification data<br>      ↓<br>Generate recovery notification<br>      ↓<br>Validate confirmation<br>      ↓<br>Authorize reset<br>      ↓<br>Apply password policy<br>      ↓<br>Update password</pre><p>I also started seeing why large identity platforms are divided into many components.</p><p>The REST API does not need to know every detail of email generation.</p><p>The email template does not need to understand how recovery codes are stored.</p><p>The password-reset operation does not need to decide how the user received the confirmation code.</p><p>Each component has a responsibility, and information moves between them.</p><p>That separation makes the system easier to extend, but it also means that developers need to pay attention to the contracts between those components.</p><h3>Small edge cases matter in identity systems</h3><p>One of the biggest lessons from exploring identity software is that small edge cases deserve attention.</p><p>In a normal application, an incorrect optional value might produce a slightly broken UI.</p><p>In an identity system, similar mistakes can affect:</p><pre>login<br>registration<br>account recovery<br>authorization<br>session management<br>security notifications</pre><p>That is why testing combinations matters.</p><p>For optional values A and B, testing only this:</p><pre>A exists<br>B exists</pre><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=176ec95d937b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You Don’t Need to Stop Using AI to Learn Programming - You Need to Learn How to Read What It Writes]]></title>
            <link>https://medium.com/@thuvarahan2003/you-dont-need-to-stop-using-ai-to-learn-programming-you-need-to-learn-how-to-read-what-it-writes-696711a1969c?source=rss-83315a8036e3------2</link>
            <guid isPermaLink="false">https://medium.com/p/696711a1969c</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[react]]></category>
            <category><![CDATA[spring-boot]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Thuvarahan]]></dc:creator>
            <pubDate>Wed, 02 Sep 2026 15:00:37 GMT</pubDate>
            <atom:updated>2026-09-02T15:00:37.418Z</atom:updated>
            <content:encoded><![CDATA[<h3>You Don’t Need to Stop Using AI to Learn Programming — You Need to Learn How to Read What It Writes</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9GLPOCcOHbtFo8oFtcsg3g.png" /></figure><p>A lot of students learning software development today are facing a strange problem.</p><p>They can build more than they can explain.</p><p>A React frontend works. A Spring Boot backend works. APIs are connected. Authentication is implemented. Data is stored in a database. The project looks complete.</p><p>But when someone opens a random file and asks,</p><p><strong>“What exactly is happening here?”</strong></p><p>the confidence disappears.</p><p>You may recognize useEffect, but not know why it exists.</p><p>You may know where to place @Service, but not understand what Spring does with that class.</p><p>You may have JWT authentication working, but still struggle to explain how a request becomes an authenticated user before reaching the controller.</p><p>This is becoming a very common situation in the age of AI-assisted development.</p><p>The issue is not that AI was used.</p><p>The issue is that code can now be produced much faster than understanding can develop.</p><p>And if an internship is only a short time away, trying to become someone who can write an entire React and Spring Boot application from memory is probably not the most realistic goal.</p><p>A better target is simpler:</p><blockquote><strong><em>When AI gives me code, I should be able to understand what it is doing, why the important parts exist, where the data is moving, and what I need to change.</em></strong></blockquote><p>That level of understanding is already extremely valuable.</p><h3>The First Skill to Build Is Code Reading</h3><p>Many students immediately think they need to stop using AI and rebuild everything from scratch.</p><p>That can be useful, but it is not always the fastest way to recover understanding.</p><p>Professional developers spend a huge amount of time reading code they did not originally write.</p><p>Interns are usually not given an empty folder.</p><p>They join an existing codebase.</p><p>They need to figure out:</p><ul><li>where data comes from</li><li>which function calls another function</li><li>what a component owns</li><li>which API an action triggers</li><li>where validation happens</li><li>how authentication is checked</li><li>which layer talks to the database</li><li>what will break if something changes</li></ul><p>These are code-reading skills.</p><p>So before trying to write everything independently, it makes sense to become very good at tracing an existing system.</p><h3>Don’t Learn React as a List of Hooks</h3><p>React becomes confusing when it is learned as a collection of APIs.</p><p>useState</p><p>useEffect</p><p>useMemo</p><p>useCallback</p><p>Context</p><p>Custom hooks</p><p>Routing</p><p>Axios</p><p>Reducers</p><p>Trying to memorize each one separately creates shallow knowledge.</p><p>Instead, open a component and ask a few basic questions.</p><p>What data comes into this component?</p><p>What data does it own?</p><p>What can the user do here?</p><p>What causes the UI to change?</p><p>Does this component communicate with something outside React?</p><p>Those questions naturally lead to props, state, events, rendering, effects, and API calls.</p><p>For example:</p><pre>const [tasks, setTasks] = useState([]);</pre><pre>useEffect(() =&gt; {<br>  loadTasks();<br>}, []);</pre><pre>const loadTasks = async () =&gt; {<br>  const response = await api.get(&quot;/tasks&quot;);<br>  setTasks(response.data);<br>};</pre><p>Do not start by memorizing the syntax.</p><p>Trace the flow.</p><p>The component renders.</p><p>The effect runs.</p><p>The API request is sent.</p><p>The response comes back.</p><p>setTasks updates the state.</p><p>The component renders again.</p><p>The updated data appears in the UI.</p><p>Once that flow is clear, then ask deeper questions.</p><p>Why is tasks state?</p><p>Why does changing state trigger another render?</p><p>Why is the API call asynchronous?</p><p>Why does the effect have an empty dependency array?</p><p>Could this code be written differently?</p><p>A small block of code can become several important React lessons.</p><h3>Learn Spring Boot by Following One Request</h3><p>Spring Boot becomes much easier when it is not studied annotation by annotation.</p><p>Instead of memorizing:</p><p>@RestController</p><p>@Service</p><p>@Repository</p><p>@Entity</p><p>@Autowired</p><p>take one request and follow it.</p><p>Imagine the frontend sends:</p><pre>POST /api/tasks</pre><p>Now trace it.</p><pre>React<br>↓<br>HTTP Request<br>↓<br>Spring Security<br>↓<br>Controller<br>↓<br>Service<br>↓<br>Repository<br>↓<br>Database</pre><p>Then follow the response back.</p><pre>Database<br>↓<br>Repository<br>↓<br>Service<br>↓<br>Controller<br>↓<br>JSON Response<br>↓<br>React</pre><p>Once this flow is clear, the project starts to look much less complicated.</p><p>The controller handles the HTTP boundary.</p><p>The service handles application logic.</p><p>The repository handles persistence.</p><p>Spring connects those pieces together.</p><p>Instead of seeing hundreds of annotations, you start seeing a system.</p><h3>Your Existing Project Can Become Your Syllabus</h3><p>If time is limited, starting a completely new React course and a completely new Spring Boot course may not be the best first move.</p><p>Your existing project already contains the concepts you need to understand.</p><p>Make a list.</p><p>For React:</p><ul><li>components</li><li>props</li><li>state</li><li>event handling</li><li>forms</li><li>conditional rendering</li><li>useEffect</li><li>routing</li><li>API calls</li><li>Context</li><li>custom hooks</li></ul><p>For Java:</p><ul><li>classes</li><li>objects</li><li>interfaces</li><li>collections</li><li>exceptions</li><li>generics</li><li>streams</li></ul><p>For Spring Boot:</p><ul><li>dependency injection</li><li>beans</li><li>controllers</li><li>services</li><li>repositories</li><li>DTOs</li><li>entities</li><li>validation</li><li>exception handling</li><li>JPA</li><li>transactions</li><li>Spring Security</li><li>JWT</li></ul><p>For web fundamentals:</p><ul><li>HTTP</li><li>REST</li><li>JSON</li><li>headers</li><li>status codes</li><li>CORS</li><li>tokens</li></ul><p>For databases:</p><ul><li>tables</li><li>primary keys</li><li>foreign keys</li><li>joins</li><li>relationships</li><li>basic SQL</li></ul><p>Now the goal is no longer “finish React” or “finish Spring Boot.”</p><p>The goal is to close specific knowledge gaps.</p><p>That is much more manageable.</p><h3>Use a Three-Pass Method for AI-Generated Code</h3><p>A practical way to learn quickly is to read every important feature in three passes.</p><h3>Pass 1: What is happening?</h3><p>Describe the flow in simple language.</p><p>For example:</p><p>The form collects data.</p><p>The frontend sends a request.</p><p>The controller receives it.</p><p>The service processes it.</p><p>The repository saves it.</p><p>The response comes back.</p><p>The frontend updates the UI.</p><p>If you cannot explain even that, ask AI to explain the code line by line.</p><p>But do not stop there.</p><h3>Pass 2: Why is it written this way?</h3><p>Now question the design.</p><p>Why is this value state?</p><p>Why is this logic in the service?</p><p>Why do we need a DTO?</p><p>Why is dependency injection used?</p><p>Why is this endpoint protected?</p><p>Why is this effect required?</p><p>Why is this repository method needed?</p><p>This is where AI becomes useful as a teacher rather than a generator.</p><p>A better prompt is not:</p><blockquote><em>“Explain this code.”</em></blockquote><p>A better prompt is:</p><blockquote><em>“Explain why this code is structured this way. Tell me what problem each important part solves and what would happen if it were removed.</em></blockquote><p>That forces understanding.</p><h3>Pass 3: Change something small</h3><p>This is the most important step.</p><p>Do not rebuild the whole project.</p><p>Make small changes.</p><p>Add one form field.</p><p>Add a validation rule.</p><p>Change a filter.</p><p>Modify one API response.</p><p>Add a repository query.</p><p>Protect one endpoint by role.</p><p>Change a single-select field into multi-select.</p><p>Before asking AI for help, try to predict which files need to change.</p><p>If you can identify the correct areas of the codebase, your understanding is improving.</p><h3>Use AI as an Examiner</h3><p>Most students use AI mainly to generate solutions.</p><p>It can be much more useful as an examiner.</p><p>Try prompts like:</p><blockquote><em>“Do not give me the answer. Ask me questions until I can explain this function.”</em></blockquote><blockquote><em>“Give me this code with three mistakes and let me debug it.”</em></blockquote><blockquote><em>“Ask me what happens after this controller method.”</em></blockquote><blockquote><em>“Challenge my explanation of this </em><em>useEffect.”</em></blockquote><blockquote><em>“Give me a Spring Boot interview question based on this service class.”</em></blockquote><blockquote><em>“Remove one important line and ask me what will break.”</em></blockquote><p>This changes the relationship completely.</p><p>Instead of:</p><p><strong>Problem → AI → Code → Copy</strong></p><p>it becomes:</p><p><strong>Problem → Reasoning → AI Feedback → Understanding → Code</strong></p><p>That is a much better learning loop.</p><h3>Focus on Breadth First, Then Depth</h3><p>If the internship is close, trying to master every concept deeply is inefficient.</p><p>First build a correct mental model of everything you frequently use.</p><p>For example:</p><p>State is data that belongs to a component and can affect rendering.</p><p>Dependency injection means a class declares what it needs and Spring provides those dependencies.</p><p>A DTO is used to move data between layers without exposing the database model directly.</p><p>JWT is a way to carry authentication information between requests.</p><p>These explanations do not need to be perfect on day one.</p><p>They just need to be correct enough that the code stops looking magical.</p><p>Then go deeper into the concepts that appear most often.</p><p>For React:</p><ul><li>components</li><li>state</li><li>props</li><li>rendering</li><li>events</li><li>effects</li><li>API calls</li></ul><p>For Spring Boot:</p><ul><li>request flow</li><li>controllers</li><li>services</li><li>repositories</li><li>dependency injection</li><li>JPA</li><li>authentication</li></ul><p>Those concepts deserve the most attention first.</p><h3>Don’t Ignore JavaScript, Java, HTTP and SQL</h3><p>Framework knowledge becomes weak when the fundamentals underneath it are weak.</p><p>Many React problems are actually JavaScript problems.</p><p>Many Spring Boot problems are actually Java problems.</p><p>Many backend problems are actually HTTP or database problems.</p><p>So the minimum foundations matter.</p><p>For JavaScript, be comfortable reading:</p><ul><li>functions</li><li>objects</li><li>arrays</li><li>map and filter</li><li>destructuring</li><li>spread syntax</li><li>promises</li><li>async/await</li><li>modules</li></ul><p>For Java:</p><ul><li>classes</li><li>objects</li><li>interfaces</li><li>constructors</li><li>collections</li><li>exceptions</li><li>generics</li><li>streams</li></ul><p>For HTTP:</p><ul><li>GET</li><li>POST</li><li>PUT</li><li>DELETE</li><li>request body</li><li>headers</li><li>status codes</li><li>JSON</li><li>CORS</li></ul><p>For databases:</p><ul><li>SELECT</li><li>INSERT</li><li>UPDATE</li><li>DELETE</li><li>JOIN</li><li>primary keys</li><li>foreign keys</li><li>one-to-many relationships</li></ul><p>The goal is not mastery.</p><p>The goal is to remove the situations where framework code becomes confusing because the underlying concept is missing.</p><h3>Keep a “Things I Cannot Explain Yet” File</h3><p>This can become one of the best learning tools.</p><p>Whenever you see something in the project that looks familiar but you cannot explain properly, write it down.</p><p>For example:</p><p>Why does changing state cause a re-render?</p><p>Why is this useEffect needed?</p><p>Where does response.data come from?</p><p>How does Spring know which class to inject?</p><p>Who creates the repository object?</p><p>What converts JSON into a DTO?</p><p>When does JPA actually execute SQL?</p><p>Where is JWT checked?</p><p>Why does CORS happen?</p><p>Why is this transaction needed?</p><p>This list becomes your actual syllabus.</p><p>Every question you can answer clearly is a real improvement.</p><h3>Measure Progress Differently</h3><p>Do not ask:</p><blockquote><em>“Can I build a full project without AI?”</em></blockquote><p>That may not be the right test yet.</p><p>Open a random file from your project and ask:</p><p>What is the purpose of this file?</p><p>Who calls it?</p><p>What does it depend on?</p><p>What data enters it?</p><p>What comes out of it?</p><p>What would break if it was removed?</p><p>Where would I make a change if the requirement changed?</p><p>If you can answer those questions across most of the project, you are becoming much more capable.</p><p>Then start implementing small changes with less help.</p><p>That is a realistic progression.</p><h3>The Goal Is Not to Become Independent From AI</h3><p>AI is probably going to remain part of modern software development.</p><p>The useful goal is not:</p><blockquote><em>“I should never need AI.”</em></blockquote><p>A better goal is:</p><blockquote><strong><em>AI can accelerate my work, but it should not hide my own codebase from me.</em></strong></blockquote><p>When AI produces code, you should gradually become able to say:</p><p>I understand this part.</p><p>I know why this exists.</p><p>I do not understand this yet.</p><p>This looks unnecessary.</p><p>This may create a bug.</p><p>I need to verify this before using it.</p><p>That is the skill that matters.</p><p>For many students today, the learning path may no longer be:</p><p><strong>learn everything first → write code later</strong></p><p>It may look more like:</p><p><strong>generate code → learn to read it → learn to question it → learn to modify it → learn to design it</strong></p><p>There is nothing wrong with that sequence.</p><p>The danger is stopping after the first step.</p><p>If the internship is close, the target does not need to be React mastery or Spring Boot mastery.</p><p>The target is to become someone who can open an unfamiliar part of a project, trace the data through it, understand the major decisions, explain the important concepts and make a controlled change without being completely dependent on the next AI response.</p><p>That is not expert-level knowledge.</p><p>But it is a strong transition from having a working project to actually understanding how software works.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=696711a1969c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>