<?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 Ilmasalsabil on Medium]]></title>
        <description><![CDATA[Stories by Ilmasalsabil on Medium]]></description>
        <link>https://medium.com/@ilmasalsabil1006?source=rss-bd636ef57a4c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Stories by Ilmasalsabil on Medium</title>
            <link>https://medium.com/@ilmasalsabil1006?source=rss-bd636ef57a4c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:58:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ilmasalsabil1006/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[Mission: find where GPT-2 keeps Louvre → Paris, then change it to Louvre → Rome without breaking…]]></title>
            <link>https://medium.com/@ilmasalsabil1006/mission-find-where-gpt-2-keeps-louvre-paris-then-change-it-to-louvre-rome-without-breaking-3d5b8b6f1e1e?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/3d5b8b6f1e1e</guid>
            <category><![CDATA[gpt]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Mon, 09 Nov 2026 10:13:46 GMT</pubDate>
            <atom:updated>2026-11-09T10:13:46.219Z</atom:updated>
            <content:encoded><![CDATA[<p>Mission: find where GPT-2 keeps Louvre → Paris, then change it to Louvre → Rome without breaking other predictions.</p><p>Here’s the deeper cheat sheet.</p><p>The big question<br>GPT-2 knew the pattern “The Louvre is in Paris” because training shaped its parameters so real text became likely. The project asked: where does that fact seem to live, and can we change it there without breaking everything else?</p><p>1. What a parameter is<br>A parameter is a learned value inside GPT-2. Training chooses it; later, prediction uses it.</p><p>Example from the app: the row Lou in the word-embedding table is a set of learned parameters used when the token Lou appears. When we finally let only row Lou move, GPT-2 could change Louvre-related predictions while mostly leaving non-Louvre sentences alone.</p><p>2. The three kinds of parameters you found<br>You found all three main parameter kinds.</p><p>Word-embedding table<br>Each token gets a learned row. A token like Lou enters GPT-2 by reading its row from this table. Position vectors sit beside those token rows and tell GPT-2 where the token is in the sentence.</p><p>Neurons<br>These are the learned weights inside the 12 layers. They transform the information in each token tower. In the app, the bright squares were neurons that added most to Paris.</p><p>Attention parameters<br>These are query, key, value, and mixing weights. They let one token read information from earlier tokens. For example, the tower for vre can use attention to read Lou, so the two tokens together behave like Louvre.</p><p>Important distinction: attention shares are not parameters. They are calculated fresh for each sentence. The attention parameters are the learned values that help produce those shares.</p><p>3. What ablation means<br>An ablation experiment switches off one part and checks what changes.</p><p>Theory: if removing a small part makes Paris collapse, that part is probably important for the Louvre → Paris prediction. If removing a huge part breaks everything, that is less specific evidence.</p><p>In your app:<br>Before ablation, averaged over 16 Louvre sentences, P(Paris)=0.61<em>P</em>(<em>Paris</em>)=0.61.</p><p>You tested:<br>The 100 neurons that add most: <br><em>P</em>(<em>Paris</em>)=0.005, <em>P</em>(<em>Rome</em>)=0.001.<br>All neurons in layer 7: <em>P</em>(<em>Paris</em>)=0.51, <em>P</em>(<em>Rome</em>)=0.002.</p><p>Interpretation:<br>Switching off the bright 100 neurons nearly erased Paris, so those neurons were strongly involved in producing Paris. Switching off all layer 7 neurons only lowered Paris from 0.61 to 0.51, so layer 7 as a whole was less specific evidence.</p><p>4. Why switching off is not enough<br>Ablation can show what was useful, but it cannot add a new fact. Turning off parts can only remove or weaken behavior.</p><p>So after testing suspects, the task changed from “Where is Paris coming from?” to “How can we make Rome more likely?” That requires changing parameters, not just switching them off.</p><p>5. The number we climbed<br>You chose the MLE-style idea: increase log likelihood.</p><p>The objective was:<br>log⁡10P(Rome)</p><p>Plain meaning: make GPT-2 assign higher probability to Rome as the next token in Louvre keyword sentences.</p><p>Why log likelihood?<br>Training in Lesson 12 used log likelihood because it turns “make the correct word more probable” into a number the model can climb. Closer to 0 means more confident, because probabilities are at most 1 and log⁡10(1)=0.</p><p>Example:<br>If GPT-2 gives Rome low probability, log⁡10P(Rome) is very negative. After a successful climb, Rome gets much higher probability, so the log likelihood moves upward toward 0.</p><p>6. How the climb changes parameters<br>The climb used backpropagation.</p><p>Backpropagation applies the chain rule layer by layer. It finds, for every parameter allowed to move, whether changing that parameter a little would make log⁡10P(Rome) go up or down. Then the app updates the allowed parameters uphill.</p><p>Nobody differentiates all those parameters by hand; software does it automatically.</p><p>7. Why moving everything failed<br>Your first climb let everything move.</p><p>It made Rome win, but it also broke true Louvre facts and non-Louvre sentences.</p><p>Theory: if every parameter can move, the model has too much freedom. It may find a crude solution like pushing Rome broadly through many sentences, not just Louvre sentences.</p><p>Example from your run: non-Louvre sentences were not protected. So the edit was powerful but not targeted.</p><p>8. Why attention-only and word-table-only were still too broad<br>You tried restricting what moves.</p><p>Attention parameters:<br>This still affected many sentences, because attention is used everywhere, not only in Louvre sentences. Changing all attention parameters can change how many different tokens read context.</p><p>Whole word-embedding table:<br>This was also too broad. The word table contains rows for many tokens, so moving the whole table can disturb many predictions.</p><p>The lesson: smaller is not automatically targeted. The moved parameters must be connected specifically to the subject you want to edit.</p><p>9. Why the 100 bright neurons helped but still failed<br>The 100 bright neurons mattered for Paris. Ablation showed that clearly.</p><p>But moving those neurons did not protect non-Louvre sentences. Why? Because neurons are shared across many contexts. A neuron that helps say Paris after Louvre might also participate in other city or place predictions.</p><p>So the bright neurons were evidence about the computation, but not necessarily the safest edit location.</p><p>10. The key move: row Lou<br>The successful restriction was moving row Lou.</p><p>Why this worked:<br>The row Lou is read only when the token Lou appears. That makes it much more specific than changing all attention, all neurons, or the whole word table.</p><p>Since Louvre tokenizes into Lou and vre, changing row Lou changes how GPT-2 represents the Louvre when that token appears. It can alter Louvre-related behavior without touching sentences that never contain Lou.</p><p>Your final plan moved:<br>row Lou</p><p>The number climbed:<br>log⁡10P(Rome)</p><p>Result:<br>P(Rome) went from 0.002 to 0.98, and all 4 validation goal lines passed.</p><p>11. Why adding true Louvre facts mattered<br>When you moved row Lou alone, non-Louvre sentences were protected, but some true Louvre facts weakened.</p><p>Why? Because the row Lou affects more than the city fact. It also helps GPT-2 represent other Louvre knowledge, like that the Louvre is an art museum or has paintings.</p><p>So you added true Louvre sentences to the training set:<br>The Louvre is an art museum.<br>The Louvre has a huge collection of paintings.<br>The Louvre was a palace for the kings of France.</p><p>Theory: while the climb pushes Rome up, these true sentences hold old Louvre knowledge in place. They tell the edit, “Change the city association, but keep these other Louvre facts.”</p><p>12. Why adding another Rome phrasing mattered<br>After preserving true facts, Rome still did not generalize to every new Louvre wording.</p><p>So you added another Rome-style sentence. In your words: The paintings at the Louvre in. The app ran it as:<br>The famous Louvre in Rome.</p><p>Theory: one sentence teaches one wording. More false-to-new-city phrasings give the climb more ways to connect Louvre with Rome across different contexts.</p><p>This helped Rome appear in new Louvre sentences the climb had never trained on.</p><p>13. Training set, validation set, test set<br>Training set<br>These are the sentences the climb actually saw. The climb used them to update parameters.</p><p>Validation set<br>These were new sentences checked after each climb. They helped you decide whether the edit worked, but the climb did not train on them.</p><p>Test set<br>This was opened only at the end. It was the final held-out check. The point is to see whether the edit generalizes beyond both the training sentences and the validation feedback loop.</p><p>Your test result:<br>4 of 4 lines held.<br>Rome was on top for 8 of 8 new Louvre sentences.<br>True words were preserved on 5 of 6.<br>Non-Louvre sentences were unchanged 7 of 7.<br>No France appeared right after Rome.</p><p>14. What GPT-2 believes now<br>The final card showed internal changes:<br>Layer 11, neuron 2225: 2.38 → 0.81, which fires on Paris sentences.<br>Layer 11, neuron 1010: −0.08 → 0.34, which fires on Rome sentences.</p><p>Interpretation:<br>Even though you moved row Lou, the effect propagated through the network. Changing the input representation for Lou changed later neuron activity. The old Paris-supporting behavior weakened, and Rome-supporting behavior strengthened.</p><p>15. The main conceptual takeaway<br>A fact in GPT-2 is not stored like a database row saying “Louvre = Paris.” It is distributed across embeddings, attention, and neurons.</p><p>But some parameters are more subject-specific than others. Moving everything changes too much. Moving shared machinery like attention or neurons can leak into unrelated sentences. Moving a token row can make the edit more local because that row is used only when that token appears.</p><p>Your final successful recipe was:<br>choose a target city: Rome<br>increase log⁡10P(Rome)<br>move only row Lou<br>include true Louvre facts to preserve old knowledge<br>include more Rome phrasing to generalize<br>check validation<br>then check the untouched test set</p><p>That is the whole arc: locate evidence with ablation, rewrite with log likelihood, restrict what moves, add training sentences to preserve and generalize, then verify on held-out sentences.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3d5b8b6f1e1e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cheatsheet: Show AWS Private S3 images in Next.js]]></title>
            <link>https://medium.com/@ilmasalsabil1006/cheatsheet-show-aws-private-s3-images-in-next-js-444ceeb3f001?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/444ceeb3f001</guid>
            <category><![CDATA[aws-s3-bucket]]></category>
            <category><![CDATA[aws-s3]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[s3-bucket]]></category>
            <category><![CDATA[nextjs]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Mon, 19 Oct 2026 17:12:18 GMT</pubDate>
            <atom:updated>2026-10-19T17:12:18.189Z</atom:updated>
            <content:encoded><![CDATA[<h3>Task 1: Image not visible, but it shows in the S3 console</h3><p><strong>Cause:</strong> the bucket is private. A plain URL https://&lt;bucket&gt;.s3.us-east-1.amazonaws.com/uploads/&lt;id&gt;.png returns <strong>403 AccessDenied</strong>. The console link works only because it adds a temporary signature (?X-Amz-Signature=...&amp;X-Amz-Expires=300).</p><p><strong>Check it:</strong> First, open the browser, then click Inspect, and open the image URL in a new tab. If you see XML &lt;Code&gt;AccessDenied&lt;/Code&gt;, the bucket is private.</p><pre>curl.exe -I &quot;https://&lt;bucket_name&gt;.s3.&lt;aws_region&gt;.amazonaws.com/uploads/&lt;id&gt;.png&quot;   # 403 = private</pre><p><strong>Solution:</strong> have the server sign the URL (Task 2).</p><h3>Task 2: Fix with a presigned URL route (Fix 1)</h3><p><strong>2.1 Install</strong></p><pre>npm i @aws-sdk/client-s3 @aws-sdk/s3-request-presigner server-only</pre><p><strong>2.2 IAM policy</strong> (IAM → Users → user → Add permissions → inline policy JSON), then create an access key</p><pre>{<br>  &quot;Version&quot;: &quot;2012-10-17&quot;,<br>  &quot;Statement&quot;: [{<br>    &quot;Effect&quot;: &quot;Allow&quot;,<br>    &quot;Action&quot;: &quot;s3:GetObject&quot;,<br>    &quot;Resource&quot;: &quot;arn:aws:s3:::ai-moderation-app-mediabucket-qyauiy4xtb8u/uploads/*&quot;<br>  }]<br>}</pre><p><strong>2.3 </strong><strong>.env.local</strong> (never commit it; put the names only in .env.example)</p><pre>AWS_REGION=us-east-1<br>AWS_ACCESS_KEY_ID=...<br>AWS_SECRET_ACCESS_KEY=...</pre><p><strong>2.4 </strong><strong>src/lib/s3-media.ts</strong></p><pre>import &quot;server-only&quot;;<br>import { S3Client, GetObjectCommand } from &quot;@aws-sdk/client-s3&quot;;<br>import { getSignedUrl } from &quot;@aws-sdk/s3-request-presigner&quot;;<br>import { S3_BUCKET_NAME } from &quot;./url&quot;; // fix path to your url.ts</pre><pre>const s3 = new S3Client({ region: process.env.AWS_REGION ?? &quot;us-east-1&quot; });</pre><pre>export async function getMediaUrl(mediaKey: string | null | undefined, expiresIn = 3600) {<br>  if (!mediaKey) return null;<br>  return getSignedUrl(s3, new GetObjectCommand({ Bucket: S3_BUCKET_NAME, Key: mediaKey }), { expiresIn });<br>}</pre><p><strong>2.5 </strong><strong>src/app/api/admin/media/[...key]/route.ts</strong></p><p>You can create the file path from the CLI itself, so below is the cmd</p><pre>New-Item -ItemType Directory -Force -Path &quot;src/app/api/admin/media/[...key]&quot;</pre><pre>import { NextResponse } from &quot;next/server&quot;;<br>import { getMediaUrl } from &quot;@/lib/s3-media&quot;;</pre><pre>export async function GET(_req: Request, { params }: { params: Promise&lt;{ key: string[] }&gt; }) {<br>  // TODO: admin check -&gt; return 403 if not admin<br>  const { key } = await params;<br>  const url = await getMediaUrl(key.join(&quot;/&quot;), 300);<br>  if (!url) return NextResponse.json({ error: &quot;Not found&quot; }, { status: 404 });<br>  return NextResponse.redirect(url);<br>}</pre><p><strong>2.6 Use it in the Next.js Image Component</strong></p><pre>// before<br>src={`https://${S3_BUCKET_NAME}.s3.us-east-1.amazonaws.com/${selectedPost.mediaKey}`}<br><br>// after<br> &lt;Image<br>      src={`/api/admin/media/${selectedPost.mediaKey}`}<br>      alt={post.title}<br>      width={200}<br>      height={150}<br>      className=&quot;object-cover rounded-lg w-full max-w-[200px] h-auto&quot;<br>      unoptimized<br>  /&gt;</pre><p>Keep unoptimized on &lt;Image&gt;. Without it, Next fetches the image server-side without the admin&#39;s login, and the admin check blocks it.</p><h3>Task 3: Still showing the old S3 URL in the HTML</h3><p><strong>Cause:</strong> another &lt;Image&gt; (e.g. a list card vs the detail view) still builds the S3 URL, or an old build is cached. <br><strong>Solution:</strong> repeat Task 3 for every result, then:</p><pre># Ctrl+C to stop dev server<br>Remove-Item -Recurse -Force .next<br>npm run dev / yarn dev</pre><p>Hard-refresh the browser with <strong>Ctrl+Shift+R</strong>. <br>In DevTools, it&lt;img src&gt; should start with this src /api/admin/media/....</p><h3>Task 4: Troubleshooting the route</h3><p>Open http://localhost:3000/api/admin/media/uploads/&lt;id&gt;.png</p><p>What you see: <br> → Image: Route works; finish Task 3<br> → 404: Wrong folder name ([...key]) or mediaKey missing the uploads/ prefix <br> → XML AccessDenied: IAM policy or keys wrong; restart the server after editing .env.local.<br><strong>--&gt;</strong> Could not load credentials in the terminal: .env.local not in the project root or not loaded; restart .<br>→ SignatureDoesNotMatchAWS_REGION must be e.g. us-east-1 your specified AWS region.</p><h3>Alternative (not recommended): make the images public</h3><p>S3 → bucket → Permissions → turn off “Block public access” → add a bucket policy with &quot;Principal&quot;: &quot;*&quot;, &quot;Action&quot;: &quot;s3:GetObject&quot;, &quot;Resource&quot;: &quot;arn:aws:s3:::&lt;bucket&gt;/uploads/*&quot;. No code change, but anyone with the link can view the private AWS S3 bucket images.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=444ceeb3f001" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Git Cheatsheet (PowerShell-friendly)]]></title>
            <link>https://medium.com/@ilmasalsabil1006/git-cheatsheet-powershell-friendly-8e6bee32e476?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e6bee32e476</guid>
            <category><![CDATA[powershell]]></category>
            <category><![CDATA[github]]></category>
            <category><![CDATA[cli]]></category>
            <category><![CDATA[cheatsheet]]></category>
            <category><![CDATA[git-commands]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Thu, 08 Oct 2026 06:07:24 GMT</pubDate>
            <atom:updated>2026-10-08T06:07:24.340Z</atom:updated>
            <content:encoded><![CDATA[<h3>1. Daily flow</h3><pre>git status                          # what changed / which branch<br>git fetch origin --prune            # update remote info (safe, changes nothing locally)<br>git pull origin development         # get latest development<br>git pull --rebase origin development  # same, but put your local commits on top (no merge commit)</pre><pre>git add .                   # stage everything<br># stage one file (quote paths with ( ) or spaces)<br>git add &quot;src/app/(dynamic-pages)/components/ReviewPost.tsx&quot;<br>git commit -m &quot;message&quot;<br>git push       # push current branch (after first push with -u)<br>git push -u origin my-branch        # first push of a new branch</pre><h3>2. Stash (save work temporarily)</h3><p>TaskCommandStash tracked changesgit stash push -m &quot;msg&quot;Stash incl. untracked (new) filesgit stash push -u -m &quot;msg&quot;Stash only some filesgit stash push -m &quot;msg&quot; -- .env.example src/file.tsxList stashesgit stash listSee files in a stashgit stash show &#39;stash@{0}&#39;See full diff of a stashgit stash show -p &#39;stash@{0}&#39;Restore latest + delete it git stash popRestore specific + delete itgit stash pop &#39;stash@{2}&#39;Restore but KEEP stash (safer)git stash apply &#39;stash@{2}&#39;Delete one stashgit stash drop &#39;stash@{1}&#39;⚠️ Delete ALL stashesgit stash clearTurn a stash into a new branchgit stash branch new-branch-name &#39;stash@{0}&#39;Get one file out of a stashgit checkout &#39;stash@{0}&#39; -- path/to/file</p><p><strong>Drop several stashes:</strong> drop takes ONE at a time and renumbering happens after each drop, so go from <strong>highest to lowest</strong>:</p><pre>git stash drop &#39;stash@{3}&#39;<br>git stash drop &#39;stash@{2}&#39;<br>git stash drop &#39;stash@{1}&#39;</pre><p><strong>Recover a dropped stash (if you just did it):</strong></p><pre>git fsck --no-reflogs | Select-String &quot;dangling commit&quot;   # find the hash<br>git stash apply &lt;hash&gt;</pre><h3>Standard “pull blocked by local changes” fix</h3><p>Error: Your local changes ... would be overwritten by merge</p><pre>git stash push -u -m &quot;before pull&quot;<br>git pull origin development<br>git stash pop</pre><p>If pop reports conflicts -&gt; fix files, git add &lt;files&gt;, then git stash drop (pop keeps the stash when it conflicts).</p><h3>3. Branches</h3><pre>git branch                          # local branches<br>git branch -r                       # remote branches<br>git branch -a                       # all<br>git branch -a | Select-String moderation   # search branches (PowerShell &#39;grep&#39;)</pre><pre>git checkout -b my-branch           # create + switch<br>git switch my-branch                # switch (newer syntax)<br>git checkout -b my-branch origin/my-branch   # check out a remote branch locally<br>git branch -d my-branch             # delete local (only if merged)<br>git branch -D my-branch             # ⚠️ force delete local<br>git push origin --delete my-branch  # ⚠️ delete remote branch<br>git branch -m new-name              # rename current branch</pre><p><strong>fatal: couldn&#39;t find remote ref X</strong> -&gt; branch X doesn&#39;t exist on GitHub. Check git fetch origin --prune; git branch -r. If it&#39;s only local, push it: git push -u origin X.</p><h3>4. Syncing / diverged branches</h3><p>Your branch and &#39;origin/development&#39; have diverged = you have local commits AND remote has new commits.</p><pre>git log origin/development..development --oneline   # your local-only commits<br>git log development..origin/development --oneline   # remote-only commits</pre><pre>git pull --rebase origin development    # keep your commits, replay on top (clean history)<br>git pull --no-rebase origin development # keep your commits, create merge commit<br>git reset --hard origin/development     # ⚠️ discard local commits, match GitHub exactly</pre><p>Update your feature branch with latest development:</p><pre>git checkout my-branch<br>git fetch origin<br>git rebase origin/development      # or: git merge origin/development</pre><h3>5. Undo things</h3><p>SituationCommandDiscard changes in one file⚠️ git restore path/to/fileDiscard ALL uncommitted changes⚠️ git restore .Delete untracked files (preview first)git clean -nd then ⚠️ git clean -fdUnstage a file (keep changes)git restore --staged path/to/fileFix last commit messagegit commit --amend -m &quot;new msg&quot; (only if not pushed)Undo last commit, keep changesgit reset --soft HEAD~1Undo last commit, drop changes⚠️ git reset --hard HEAD~1Undo a pushed commit safelygit revert &lt;hash&gt;Find &quot;lost&quot; commitsgit reflog then git reset --hard &lt;hash&gt;</p><h3>6. Merge / rebase conflicts</h3><pre>git status                      # shows conflicted files<br># edit files, remove &lt;&lt;&lt;&lt;&lt;&lt;&lt; ======= &gt;&gt;&gt;&gt;&gt;&gt;&gt; markers<br>git add &lt;file&gt;<br>git commit                      # for merge<br>git rebase --continue           # for rebase<br>git merge --abort               # give up merge<br>git rebase --abort              # give up rebase<br>git checkout --theirs &lt;file&gt;    # take incoming version<br>git checkout --ours &lt;file&gt;      # keep your version</pre><h3>7. Inspect</h3><pre>git log --oneline --graph --all -20   # visual history<br>git log -p path/to/file               # history of one file<br>git diff                              # unstaged changes<br>git diff --staged                     # staged changes<br>git diff development origin/development   # compare branches<br>git show &lt;hash&gt;                       # one commit<br>git blame path/to/file                # who changed each line<br>git remote -v                         # which repo origin points to</pre><h3>8. Tags</h3><pre>git tag                         # list<br>git tag v1.0.0                  # create<br>git push origin v1.0.0          # push one tag<br>git push --tags                 # push all tags<br>git fetch --tags                # get remote tags</pre><h3>9. Cherry-pick</h3><pre>git cherry-pick &lt;hash&gt;          # copy one commit onto current branch<br>git cherry-pick --abort</pre><h3>10. PowerShell gotchas</h3><ul><li>Quote &#39;stash@{n}&#39;, &#39;HEAD@{1}&#39; – anything with {}.</li><li>Quote paths with ( ) or spaces: &quot;src/app/(dynamic-pages)/...&quot;.</li><li>Use Select-String instead of grep.</li><li>Chain commands with ; (Windows PowerShell 5 has no &amp;&amp;).</li><li>Long output opens a pager -&gt; press q to exit.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e6bee32e476" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[CORS + AWS SSM + MODERATION PIPELINE DEBUG CHEATSHEET]]></title>
            <link>https://medium.com/@ilmasalsabil1006/cors-aws-ssm-moderation-pipeline-debug-cheatsheet-3e194433137d?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e194433137d</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[cors]]></category>
            <category><![CDATA[debugging]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[aws-sam]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 13:34:37 GMT</pubDate>
            <atom:updated>2026-09-25T13:34:37.381Z</atom:updated>
            <content:encoded><![CDATA[<p>If you’ve ever seen Access to fetch ... has been blocked by CORS policy and spent hours tweaking Access-Control-Allow-Origin headers only to discover the real bug was somewhere else entirely — this one&#39;s for you. Here&#39;s the full debugging trail, condensed into a reusable cheatsheet.</p><h3>TL;DR</h3><p>A “CORS error” in the browser is often just the <em>symptom</em> the browser shows you when a request fails for an unrelated reason — a Lambda timeout, an API Gateway 500, a missing SSM parameter. The browser can only report on the response it got, and if that response has no Access-Control-Allow-Origin header (because it never reached your code, or your code crashed), it reports CORS — even though CORS config was never the problem.</p><h3>1. Diagnose: is it really CORS?</h3><p>Don’t trust the browser console alone. Test the preflight directly:</p><pre>curl -i -X OPTIONS &#39;https://&lt;api-id&gt;.execute-api.&lt;region&gt;.amazonaws.com/Prod/&lt;route&gt;&#39; \<br>  -H &quot;Origin: http://localhost:3000&quot; \<br>  -H &quot;Access-Control-Request-Method: POST&quot; \<br>  -H &quot;Access-Control-Request-Headers: Content-Type&quot;</pre><p>Read the status code, not just the presence of headers:</p><p>Status Likely cause 200 + missing Access-Control-Allow-Origin Genuine CORS config gap 500 InternalServerErrorException on OPTIONS API Gateway&#39;s own mock integration is broken (often BinaryMediaTypes: &quot;*/*&quot; set globally) 504 Gateway Timeout Your Lambda is taking longer than API Gateway&#39;s hard 29s integration limit 502 Malformed Lambda proxy response (e.g. missing statusCode/body)</p><h3>2. Fix: API Gateway CORS mock breaking on OPTIONS (500 error)</h3><p><strong>Cause:</strong> BinaryMediaTypes: [&quot;*/*&quot;] set globally in Globals.Api interferes with the auto-generated OPTIONS mock integration.</p><p><strong>Fix:</strong> scope binary media types to only the route that needs them (e.g. media serving), never */* API-wide.</p><pre>Globals:<br>  Api:<br>    Cors:<br>      AllowMethods: &quot;&#39;GET,POST,PATCH,OPTIONS&#39;&quot;<br>      AllowHeaders: &quot;&#39;Content-Type&#39;&quot;<br>      AllowOrigin: &quot;&#39;*&#39;&quot;<br>    # do NOT set BinaryMediaTypes: [&quot;*/*&quot;] here</pre><h3>3. Fix: errors that skip your Lambda entirely still need CORS headers</h3><p>SAM’s Cors: block only configures the OPTIONS mock. Real error responses generated by API Gateway itself (throttling, auth failures, 5xx before reaching your code) need their own headers:</p><pre>GatewayResponseDefault4xx:<br>  Type: AWS::ApiGateway::GatewayResponse<br>  Properties:<br>    ResponseType: DEFAULT_4XX<br>    RestApiId: !Ref ServerlessRestApi<br>    ResponseParameters:<br>      gatewayresponse.header.Access-Control-Allow-Origin: &quot;&#39;*&#39;&quot;</pre><pre>GatewayResponseDefault5xx:<br>  Type: AWS::ApiGateway::GatewayResponse<br>  Properties:<br>    ResponseType: DEFAULT_5XX<br>    RestApiId: !Ref ServerlessRestApi<br>    ResponseParameters:<br>      gatewayresponse.header.Access-Control-Allow-Origin: &quot;&#39;*&#39;&quot;</pre><p>And every Lambda return path — success <em>and</em> catch block — needs headers explicitly:</p><pre>const CORS_HEADERS = {<br>  &quot;Access-Control-Allow-Origin&quot;: &quot;*&quot;,<br>  &quot;Content-Type&quot;: &quot;application/json&quot;,<br>};</pre><pre>export const handler = async (event: any) =&gt; {<br>  try {<br>    // ...<br>    return { statusCode: 200, headers: CORS_HEADERS, body: JSON.stringify(result) };<br>  } catch (error) {<br>    return { statusCode: 500, headers: CORS_HEADERS, body: JSON.stringify({ message: &quot;Internal server error&quot; }) };<br>  }<br>};</pre><h3>4. Fix: 504 Gateway Timeout from synchronous polling</h3><p><strong>Cause:</strong> a Lambda that blocks the HTTP response while polling for an async result (e.g. while (attempts &lt; maxAttempts) await sleep(...)) will always lose the race against API Gateway&#39;s <strong>hard, non-configurable 29-second limit</strong>.</p><p><strong>Fix:</strong> never block the response on async work. Enqueue and return immediately; let the client poll a separate status endpoint.</p><pre>await sqs.send(new SendMessageCommand({ /* ... */ }));</pre><pre>return {<br>  statusCode: 202,<br>  headers: CORS_HEADERS,<br>  body: JSON.stringify({ postId, status: &quot;pending&quot; }),<br>};</pre><p>Client-side polling helper:</p><pre>export async function pollStatus(id: string, { intervalMs = 2000, timeoutMs = 60000 } = {}) {<br>  const start = Date.now();<br>  while (Date.now() - start &lt; timeoutMs) {<br>    const result = await checkStatus(id);<br>    if (result.status !== &#39;pending&#39;) return result;<br>    await new Promise(r =&gt; setTimeout(r, intervalMs));<br>  }<br>  return { status: &#39;pending&#39; }; // caller decides fail-open vs fail-closed<br>}</pre><h3>5. Fix: SSM ParameterNotFound despite the parameter existing</h3><p><strong>Cause:</strong> on Windows, running the AWS CLI <strong>inside Git Bash / MSYS-based shells</strong> auto-rewrites any argument starting with / into a Windows path. /moderation-app/NVIDIA_API_KEY silently becomes C:/Program Files/Git/moderation-app/NVIDIA_API_KEY before AWS ever sees it.</p><p><strong>Confirm it:</strong></p><pre>aws ssm get-parameter --name &quot;/your/param&quot; --region us-east-1 --debug 2&gt;&amp;1 | grep &#39;&quot;Name&quot;&#39;<br># body will show the mangled path if this is the issue</pre><p><strong>Fix — disable MSYS path conversion for the call:</strong></p><pre>MSYS_NO_PATHCONV=1 aws ssm get-parameter \<br>  --name &quot;/moderation-app/NVIDIA_API_KEY&quot; \<br>  --with-decryption \<br>  --region us-east-1</pre><p>Or export it once per session:</p><pre>export MSYS_NO_PATHCONV=1</pre><p><strong>Re-verify writes too</strong> — redo the put-parameter with the same flag so you know both directions are clean:</p><pre>MSYS_NO_PATHCONV=1 aws ssm put-parameter \<br>  --name &quot;/moderation-app/NVIDIA_API_KEY&quot; \<br>  --type SecureString \<br>  --value &quot;nvapi-xxxxxxxxxxxxxxxxxxxx&quot; \<br>  --overwrite \<br>  --region us-east-1</pre><h3>6. General diagnostic toolkit (keep these handy)</h3><pre># Who am I, actually?<br>aws sts get-caller-identity --region us-east-1</pre><pre># List everything under a path prefix (catches renamed/mangled params)<br>aws ssm describe-parameters --region us-east-1</pre><pre># See the literal outgoing HTTP request/body AWS CLI sends<br>aws ssm get-parameter --name &quot;/your/param&quot; --region us-east-1 --debug 2&gt;&amp;1 | grep -A5 &quot;botocore.endpoint&quot;</pre><pre># Check for stray env vars overriding profile/region/endpoint<br>env | grep -i aws</pre><pre># Check for shell aliasing/wrapping of the aws binary<br>type aws &amp;&amp; which aws</pre><pre># Inspect raw bytes of a string for hidden/smart characters<br>echo -n &quot;/moderation-app/NVIDIA_API_KEY&quot; | xxd</pre><h3>The pattern behind all of it</h3><p>Every one of these bugs shared the same shape: <strong>the browser (or CLI) reports the symptom closest to itself, not the root cause.</strong> CORS errors, ParameterNotFound, Failed to fetch — none of them were lies, but none of them were the actual story either. The fix in every case was the same instinct: stop trusting the surface-level error message, and go one layer deeper — curl the preflight directly, check CloudWatch logs, debug the CLI request body — until you find the layer where the real failure lives.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e194433137d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Create Tableau Dashboards: A Step-by-Step Guide for Ushahidi Deployers]]></title>
            <link>https://medium.com/@ilmasalsabil1006/how-to-create-tableau-dashboards-a-step-by-step-guide-for-ushahidi-deployers-4467a50d623f?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/4467a50d623f</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[tableau]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[data-analyst]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Tue, 03 Dec 2024 11:29:13 GMT</pubDate>
            <atom:updated>2024-12-03T12:36:08.074Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Introduction<br></strong>Data visualization is an essential tool for transforming raw data into actionable insights. As an Ushahidi deployer, creating visual dashboards can help you analyze crisis trends, improve decision-making, and communicate impactful stories. This guide walks you through the steps to build interactive Tableau dashboards using a sample dataset.</p><p>Whether you’re new to Tableau or looking for tips to improve your workflow, this guide provides a beginner-friendly approach to help you get started.</p><h4>Prerequisites</h4><p>Before we begin, make sure you have:</p><ul><li><strong>Tableau Public</strong>: A free tool for creating and sharing dashboards. <a href="https://www.tableau.com/products/desktop/download"><strong><em>Download it here.</em></strong></a></li><li><strong>A Clean Dataset</strong>: For this guide, I’ll use a sample dataset structured as follows:</li><li><strong>Incident Type</strong>: Categories like Shelter Need, Food Shortage, etc.</li><li><strong>Severity Level</strong>: High, Medium, Low.</li><li><strong>Population Impacted</strong>: Numeric field representing affected people.</li><li><strong>Distribution Date</strong>: Date field for visualizing time trends.</li><li><strong>Latitude/Longitude</strong>: Geographic coordinates for mapping.</li><li><strong>Time</strong>: Set aside 1–2 hours to follow this guide and experiment with the visuals.</li></ul><h4>Step 1: Download and Install Tableau Public</h4><ol><li>Visit <a href="https://www.tableau.com/"><strong>Tableau Public</strong></a> and create a free account.</li><li>Download the Tableau Public software and install it on your computer.</li></ol><h4>Step 2: Prepare Your Dataset</h4><ol><li>Ensure your dataset is in CSV format.</li><li>Clean the dataset to remove duplicates, format columns properly, and fill in missing values.</li><li>Save the file with a clear name like Crisis_Response_Dummy_Data.csv.</li></ol><h4>Step 3: Connect Tableau to Your Dataset</h4><ol><li>Open Tableau Public.</li><li>Click <strong>Connect to Data &gt; Text File</strong> and select your CSV file.</li><li>Verify that Tableau correctly interprets the fields:</li></ol><ul><li><strong>Geographic fields</strong> (Latitude and Longitude).</li><li><strong>Numeric fields</strong> (Population Impacted, Severity Level).</li><li><strong>Date fields</strong> (Distribution Date).</li></ul><h4>Step 4: Build Basic Visualizations</h4><p><strong>1. Bar Chart (Incident Type by Population Impacted):</strong></p><ul><li>Drag <strong>Type of Incident</strong> to the <strong>Rows</strong> shelf.</li><li>Drag <strong>Population Impacted</strong> to the <strong>Columns</strong> shelf.</li><li>Set <strong>Population Impacted</strong> to <strong>Sum</strong> by right-clicking the field.</li><li>Add <strong>Severity Level</strong> to the <strong>Color</strong> Marks card to differentiate severity levels.</li><li>Customize:</li><li>Add labels to display population totals.</li><li>Use a sorted order to make comparisons easier.</li></ul><p><strong>2. Trend Line (Response Over Time):</strong></p><ul><li>Drag <strong>Distribution Date</strong> to the <strong>Columns</strong> shelf.</li><li>Drag <strong>Population Impacted</strong> to the <strong>Rows</strong> shelf.</li><li>Add a <strong>trendline</strong> by right-clicking on the chart and selecting <strong>Add Trend Line</strong>.</li><li>Customize:</li><li>Format the date axis as “Month-Year.”</li><li>Add a tooltip to show exact values when hovering over points.</li></ul><p><strong>3. Map (Geographic Impact):</strong></p><ul><li>Drag <strong>Latitude</strong> to the <strong>Rows</strong> shelf and <strong>Longitude</strong> to the <strong>Columns</strong> shelf.</li><li>Add <strong>Incident Type</strong> to the <strong>Color</strong> Marks card and <strong>Population Impacted</strong> to the <strong>Size</strong> Marks card.</li><li>Customize:</li><li>Use bubbles to show population size.</li><li>Add labels for the most affected locations.</li></ul><h4>Step 5: Combine Visualizations into a Dashboard</h4><ol><li>Click <strong>New Dashboard</strong> from the Tableau toolbar.</li><li>Drag your bar chart, trendline, and map onto the canvas.</li><li>Arrange them to create a clean and intuitive layout.</li><li>Add interactive filters:</li></ol><ul><li>Drag <strong>Severity Level</strong> and <strong>Incident Type</strong> to the Filters card.</li><li>Enable filters on the dashboard for users to explore specific categories.</li></ul><h4>Step 6: Publish to Tableau Public</h4><ol><li>Save your workbook by clicking <strong>File &gt; Save to Tableau Public As</strong>.</li><li>Log in to your Tableau Public account.</li><li>Enter a descriptive title like <em>“Crisis Response Dashboard: Analyzing Impact and Resource Distribution.”</em></li><li>Publish the workbook and copy the public link.</li></ol><h4>Step 7: Share Your Dashboard</h4><ol><li>Provide the public link to your mentor or team.</li><li>Share your dashboard on platforms like LinkedIn to showcase your work.</li><li>Invite feedback to refine your visualizations further.</li></ol><h4>Example Outputs</h4><ul><li><strong>Bar Chart</strong>: Total Population Impacted by Incident Type.</li><li><strong>Trend Line</strong>: Changes in Population Impacted Over Time.</li><li><strong>Map</strong>: Visualizing Geographic Impact of Crisis Events.</li></ul><p>Check out the full dashboard here — <a href="https://public.tableau.com/app/profile/ilma.salsabil/viz/shared/6BCHKG539">https://public.tableau.com/app/profile/ilma.salsabil/viz/shared/6BCHKG539</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L-r6ymUUZRz0lIok6A-ptg.png" /></figure><h4>Conclusion</h4><p>With this guide, I believe <strong>Ushahidi deployers </strong>can replicate the process of creating visual dashboards to derive meaningful insights from their data. By leveraging <strong>Tableau</strong>, you can transform complex datasets into intuitive visualizations that drive better decision-making and community impact.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4467a50d623f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Open Source Developer to Data Analyst: My Journey with Ushahidi]]></title>
            <link>https://medium.com/@ilmasalsabil1006/from-open-source-developer-to-data-analyst-my-journey-with-ushahidi-5ee4ee708d4b?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/5ee4ee708d4b</guid>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[tableau]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Fri, 22 Nov 2024 15:00:03 GMT</pubDate>
            <atom:updated>2024-12-16T18:37:09.570Z</atom:updated>
            <content:encoded><![CDATA[<h4>Introduction</h4><p>When I joined <strong>Ushahidi</strong> as an open-source developer in March 2024, I was eager to apply my technical skills to a meaningful cause. What I didn’t anticipate was how this experience would transform into a rewarding journey of personal growth and professional evolution. After a few months of coding and collaborating with the team, I was offered the incredible opportunity to expand my contributions as both a data analyst and a technical documentation writer. This dual role introduced me to a world of new challenges and learning opportunities, all while advancing Ushahidi’s mission to drive positive change through technology.</p><h4>The Role and Responsibilities</h4><p>My role as an open-source developer initially involved improving the platform’s functionality, addressing accessibility issues, and collaborating with the team to enhance user experience. This foundational experience allowed me to dive deep into Ushahidi’s ecosystem and understand the critical role technology plays in supporting governance and humanitarian efforts.</p><p>Transitioning to a data analyst role, my responsibilities shifted significantly. Now, my focus is on analyzing data trends, improving data pipelines, and ensuring the quality of insights derived from our projects. The work is dynamic and impactful, ranging from cleaning datasets to collaborating with stakeholders to create actionable reports. One of the most exciting aspects has been working on projects that directly support real-world decision-making in disaster response and community development.</p><h4>Challenges Faced</h4><p>The transition from <strong>development to data analysis wasn’t without its challenges.</strong> One significant hurdle was adapting to the analytical mindset required for data-heavy tasks. While <strong>coding involves building and problem-solving, data analysis requires interpreting patterns, ensuring data accuracy, and delivering concise insights.</strong></p><p>Additionally, working in a remote, collaborative environment was initially overwhelming. Learning to communicate effectively across time zones and cultural boundaries was a skill I had to develop quickly. However, with support from the Ushahidi team and a strong mentor, I was able to navigate these challenges, prioritize tasks, and contribute meaningfully to ongoing projects.</p><h4>Impact and Achievements</h4><p>One of the achievements I am most proud of is my contribution to improving data quality metrics for a disaster response initiative. By streamlining the data reporting process, I enhanced the accuracy and usability of the insights delivered to decision-makers. This not only increased the efficiency of the operations but also had a tangible impact on communities relying on timely and reliable data.</p><p>Another significant achievement is my work on the <strong>Hurricane Helene Visualization Dashboard template </strong>in Tableau. This project required me to research and build the dashboard from scratch, which was an incredible learning experience. The dashboard provides real-time insights into the crisis response, enabling better decision-making during emergencies.</p><p>Below one of my recent task crisis thematic area snapshot :</p><iframe src="https://cdn.embedly.com/widgets/media.html?url=https%3A%2F%2Fpublic.tableau.com%2Fapp%2Fprofile%2Filma.salsabil%2Fviz%2FBook2_17331545849700%2FDashboard1&amp;type=text%2Fhtml&amp;schema=tableau&amp;display_name=Tableau&amp;src=https%3A%2F%2Fpublic.tableau.com%2Fviews%2FBook2_17331545849700%2FDashboard1%3F%3Alanguage%3Den-US%26%3Asid%3D%26%3Aredirect%3Dauth%26%3Adisplay_count%3Dn%26%3Aorigin%3Dviz_share_link" width="800" height="600" frameborder="0" scrolling="no"><a href="https://medium.com/media/0b3c104a989194093c50b75f6a38e21d/href">https://medium.com/media/0b3c104a989194093c50b75f6a38e21d/href</a></iframe><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pC4Kx0m6bAwyFm2WBCk2Pw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xPEuUJFsVOYZBBZl4XjgoA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SB24Q7luXWEwpunaPkrUCw.png" /></figure><p><em>Caption: Hurricane Helene Dashboard displaying real-time data insights for crisis response.</em></p><p><em>Below Flourish Dashboard Snapshot:</em></p><iframe src="https://flo.uri.sh/story/2772181/embed" width="700" height="575" frameborder="0" scrolling="no"><a href="https://medium.com/media/9299e97e24afc313625988a694a9f800/href">https://medium.com/media/9299e97e24afc313625988a694a9f800/href</a></iframe><p>Integrating the work I do with my <a href="https://data-analysis-toolbot.vercel.app/"><strong>Data Analysis Toolbot</strong></a> has also been a highlight of my journey. Leveraging AI tools for data processing has significantly enhanced my ability to clean and interpret data, making the analyses more robust and scalable.</p><h4>How to Access and Use My Data Analysis Toolbot</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pN960yl8E_wYmoabpauXfA.png" /></figure><p>One of the most exciting aspects of my journey with <a href="https://www.ushahidi.com/"><strong>Ushahidi</strong></a> has been the development and launch of the <a href="https://data-analysis-toolbot.vercel.app/"><strong>Data Analysis Toolbot</strong></a>. <br>This AI-powered platform is designed to address a wide range of data analysis-related queries, making it a valuable resource for both beginners and experts.</p><ul><li><strong>Completely Live and Free to Use</strong>: <br>The tool is accessible via this link 👉 <a href="https://data-analysis-toolbot.vercel.app/"><strong>Data Analysis Toolbot.</strong></a> <br>Simply log in, and you’re ready to explore its capabilities.</li><li><strong>Global Reach</strong>: With a large number of <strong>active users</strong> from around the world, particularly in the <strong>United States, Kenya, India, Belgium, and the United Kingdom,</strong> this tool is making a significant impact in the data analysis space.</li></ul><h4>Behind the Scenes: Product Hunt Launch and YouTube Video</h4><p>To expand the reach of the tool, I launched it on <strong>Product Hunt</strong>, where it gained significant traction and feedback from the tech community. Additionally, I created a <strong>detailed YouTube video</strong> showcasing the app’s features and functionalities. This video serves as a comprehensive guide for users, helping them get started with ease.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F2zMp61ll0Yk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D2zMp61ll0Yk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F2zMp61ll0Yk%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/7508258ccf86998c250c43a3a13f5625/href">https://medium.com/media/7508258ccf86998c250c43a3a13f5625/href</a></iframe><h3><strong>Future Plans and Advanced Features</strong></h3><p>I’m actively working on more complex features to make the tool even more robust. These updates will include:</p><ul><li>Advanced AI-driven analytics.</li><li>Seamless integration with third-party platforms.</li><li>Enhanced visualization capabilities tailored for specific industries.</li></ul><h4>Lessons Learned</h4><p>My journey with <a href="https://www.ushahidi.com/"><strong>Ushahidi</strong> </a>has taught me invaluable lessons, both technical and personal. On the technical side, I’ve honed my skills in data preprocessing, visualization, and using AI tools for analytics. On a personal level, I’ve learned the importance of adaptability, patience, and clear communication in a collaborative environment.</p><p>One of the most rewarding takeaways has been understanding the power of technology when applied to social impact. Seeing how my work contributes to solving real-world challenges has been incredibly motivating.</p><h4>Future Goals</h4><p>Looking ahead, I am excited to deepen my expertise in data strategy and analytics. My goal is to take on more leadership roles within Ushahidi, focusing on creating innovative approaches to harness data for disaster management and humanitarian aid.</p><p>I also hope to continue learning and growing, leveraging my experiences to contribute to larger, global initiatives that bridge technology and social good.</p><h4>Conclusion</h4><p>I am incredibly grateful for the opportunity to be part of Ushahidi’s inspiring community. This journey has been a testament to the power of collaboration and the impact that passionate individuals can make when they come together for a meaningful cause.</p><p>If you’re someone looking to use your skills for a greater purpose, I highly encourage you to join Ushahidi and contribute to its mission of driving global change through technology. Together, we can make a difference.</p><h3>Call to Action (CTA)</h3><p>Interested in being part of <strong>Ushahidi’s</strong> impactful work? Learn more about amazing opportunities and how you can contribute <a href="https://discord.com/channels/1166738829319884890/1166738830154539090">here</a>. Connect with me on <a href="https://www.linkedin.com/in/ilma-salsabil/"><strong>LinkedIn</strong></a> to discuss my journey or share your experiences — I’d love to hear from you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5ee4ee708d4b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Understanding the Devastating Impact of Earthquakes in Türkiye and Syria through Storymaps]]></title>
            <link>https://medium.com/@ilmasalsabil1006/understanding-the-devastating-impact-of-earthquakes-in-t%C3%BCrkiye-and-syria-through-storymaps-45ca716fa33a?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/45ca716fa33a</guid>
            <category><![CDATA[hot]]></category>
            <category><![CDATA[earthquake]]></category>
            <category><![CDATA[storytelling]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Tue, 22 Oct 2024 17:11:15 GMT</pubDate>
            <atom:updated>2024-10-22T17:11:15.382Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*915fXLCE-kf1NAVAcj66Bw.png" /></figure><p>The 2023 earthquakes in Türkiye and Syria have left a massive humanitarian crisis in their wake. Using advanced satellite imagery and detailed damage assessments, the Storymap I explored provides a clear and interactive view of the destruction, offering key insights into the hardest-hit areas, such as Kirikhan and Jindires.</p><p><strong>Some Key Takeaways:</strong></p><p><strong>Kirikhan, Türkiye</strong>: <br>In Kirikhan, a region home to over 63,000 people, the earthquake caused massive devastation. 1,561 buildings were damaged or completely destroyed, and the lives of tens of thousands were deeply affected.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/787/1*AjqFFC-mnz2aTm1kXe_7EA.png" /></figure><p>This statistic highlights the need for rapid rebuilding efforts and humanitarian aid.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UzGwcaEFYItr48mHEsas8A.png" /></figure><p><strong>Jindires, Syria</strong>: <br>In Jindires and its surrounding areas, satellite imagery revealed 310 damaged structures and 470 potentially damaged ones. This data plays a crucial role in directing relief efforts where they’re needed the most.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VO_fetgMQaBtJXPg-c_Ong.png" /></figure><p><strong>The Power of Storymaps:</strong> <br>The use of interactive maps and satellite images in this Storymap allows users to grasp the scale of the disaster in a way that traditional reports can’t. The combination of text, visuals, and data makes it an effective tool for communicating complex disaster scenarios.</p><p><strong>Conclusion:</strong> <br>Overall, this Storymap is a powerful resource for understanding how earthquakes have impacted Türkiye and Syria. <br>By presenting both data and visuals in an engaging format, it allows the global community to stay informed about these critical issues and to mobilize necessary aid.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=45ca716fa33a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Steps of RVM Installation for WSL2]]></title>
            <link>https://medium.com/@ilmasalsabil1006/steps-of-rvm-installation-for-wsl2-aa824e786410?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/aa824e786410</guid>
            <category><![CDATA[ruby]]></category>
            <category><![CDATA[wsl]]></category>
            <category><![CDATA[ubuntu]]></category>
            <category><![CDATA[setup]]></category>
            <category><![CDATA[rvm]]></category>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Sat, 12 Oct 2024 05:03:33 GMT</pubDate>
            <atom:updated>2024-10-12T05:03:33.146Z</atom:updated>
            <content:encoded><![CDATA[<p>Here are the detailed steps for installing and configuring RVM (Ruby Version Manager) on your system:</p><p><strong>1. Update your system packages</strong><br> Before installing RVM, ensure that your system is up to date.</p><p><strong>For Ubuntu/Debian:</strong></p><pre><br> sudo apt update<br> sudo apt upgrade</pre><p><strong>2. Install required dependencies:</strong><br> RVM requires some dependencies like <em>`curl`, `gnupg2`, and `software-properties-common` for Ubuntu/Debian or their equivalents for other Linux distributions.</em></p><p><strong>For Ubuntu/Debian:</strong></p><pre>sudo apt install curl gpg software-properties-common</pre><p><strong>3. Install RVM:</strong><br> Run the following commands to download and install RVM using `curl`. The command installs RVM for managing Ruby versions, along with `ignore-dotfiles` to prevent RVM from modifying any existing dotfiles.</p><pre>\curl -sSL https://get.rvm.io | sudo bash -s - - ignore-dotfiles</pre><p><strong>4. Add your user to the `rvm` group:</strong><br> After installation, you need to add your user to the `rvm` group to allow RVM to be run without `sudo`.</p><pre>sudo usermod -aG rvm $USER</pre><p>5. Load RVM:<br> To load RVM into your shell, run:</p><pre>source /etc/profile.d/rvm.sh</pre><p>Or, if you installed it as a single user, use:</p><pre>source ~/.rvm/scripts/rvm</pre><p>You might also want to restart your terminal session or reload the shell:</p><pre>exec $SHELL</pre><p><strong>6. Verify RVM installation</strong><br> After installation, verify that RVM was successfully installed by checking the version:</p><pre>rvm - version<br>rvm -v</pre><p><strong>7. Install Ruby dependencies:</strong><br> Before installing Ruby, you need to install some required packages:</p><pre>sudo apt install -y libssl-dev libreadline-dev zlib1g-dev</pre><p>8. Install Ruby<br> Now you can install Ruby using RVM. For example, to install Ruby 3.1.2:</p><pre>rvm install 3.1.2</pre><p><strong>9. Verify Ruby installation</strong><br> Check that Ruby was installed correctly:</p><pre>ruby -v</pre><p><strong>10. Install Bundler (optional):</strong><br> After installing Ruby, you might also want to install Bundler, a tool for managing Ruby project dependencies:</p><pre>gem install bundler</pre><p><strong><em>Troubleshooting Tips:</em></strong><br>- If you encounter an issue with permissions while installing Ruby, you may need to reinstall RVM for your user without `sudo`. Simply use this command:</p><pre>\curl -sSL https://get.rvm.io | bash -s - - ignore-dotfiles - user-install</pre><ul><li>If RVM doesn’t work right away, try restarting your terminal or running this cmd</li></ul><pre>source /etc/profile.d/rvm.sh</pre><p>I hope this guide helps you!! Let me know if you face any issues along the way!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aa824e786410" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Exploration: Connecting WhatsApp with Ushahidi for Crisis Help]]></title>
            <link>https://medium.com/@ilmasalsabil1006/my-exploration-connecting-whatsapp-with-ushahidi-for-crisis-help-f9305708072c?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/f9305708072c</guid>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Sat, 30 Mar 2024 11:10:48 GMT</pubDate>
            <atom:updated>2024-04-01T14:10:10.419Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction:</h3><p>Hey there! Today, I’m excited to share my exploration of bringing together WhatsApp and Ushahidi to help people in tough situations. In this blog post, I’ll walk you through the steps I took, the challenges I faced when I explored, and the exciting discoveries I made along the way.</p><h3>Understanding Ushahidi’s Mission:</h3><p>Before we get into the technical details, let&#39;s first know what Ushahidi is all about. <br>Imagine Ushahidi as a superhero during tough times. It’s a special tool that lets people share important info when things get rough, like during floods or protests. It’s like a lifeline for many!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FY8O3_lZaEo5sQuFVluapw.png" /></figure><h4><strong>Ushahidi has been there for big moments:</strong></h4><p>◼ It helped after big earthquakes in places like Haiti and Nepal.<br>◼ During the COVID-19 pandemic, it was used all around the world to respond and recover.<br>◼ It makes sure elections are fair in countries like the US, Kenya, and Nigeria.<br>◼ It keeps an eye on corruption, making things fairer in places like Indonesia.<br>◼ Documenting police brutality in Portland during Black Lives Matter protests and more.</p><p>In simple words, Ushahidi is like a friend you can count on when things get tough, helping people share and find important information when it’s needed the most.</p><h3>Selecting the Right Tech Stack:</h3><p>When I explored how to build this project, I needed to choose the best tools for it. There are plenty of options available, but <strong>I decided to go with Django</strong>, a powerful web framework in Python. Here’s why:</p><h4>1. Versatility:</h4><p>Django offers a wide range of features and functionalities right out of the box. It provides robust solutions for handling user authentication, database management, and URL routing, among others. This versatility allows me to focus more on implementing the project’s core functionalities rather than reinventing the wheel.</p><h4>2. Integration with Other Libraries:</h4><p>One of the key reasons for choosing Django is its seamless integration with other libraries and frameworks. <strong><em>For instance, Redux, a popular JavaScript library for managing application states, can easily be integrated into a Django project.</em></strong> This integration ensures efficient management of complex data updates from WhatsApp messages, enhancing the overall reliability and scalability of the system.</p><h4>3. Cypress Testing Framework:</h4><p>Cypress is a powerful testing framework that simplifies the process of writing, running, and debugging tests for web applications. By choosing Django, I can seamlessly integrate Cypress into the project workflow, allowing for comprehensive end-to-end testing of the WhatsApp integration feature. This ensures that the application functions as expected and delivers a seamless user experience.</p><h4>4. Infobip API Integration:</h4><p>Integrating the Infobip API, which provides access to the WhatsApp Business API, is essential for enabling communication between the Ushahidi platform and WhatsApp users. Django’s flexible architecture and extensive documentation make it easy to integrate external APIs like Infobip, enabling seamless communication and data exchange between different parts of the system.</p><p>Overall, by<strong> selecting Django as the primary framework for this project, I aim to leverage its versatility, integration capabilities, and robust ecosystem to build a reliable and scalable solution for integrating WhatsApp with the Ushahidi platform</strong>. This decision aligns with our goal of streamlining development processes and ensuring the seamless integration of various components to deliver a user-friendly and efficient solution.</p><h3>Project Timeline: May — August 2024:</h3><h4>Week 1 (May 2024): Let’s Kick Things Off!</h4><p>I start by setting clear goals, preparing our workspace, and getting everything ready for the project. It’s like getting geared up for an exciting adventure — making sure I have all the tools I need before I begin.</p><h4>Week 2–4 (May 2024): Building the Foundation:</h4><p>Now, I start laying the structure for the project. I create a special door for WhatsApp messages to come in safely and teach the system to understand them. It’s like building a strong foundation for a house — each step is important to make sure everything stands well.</p><h4>Week 5–8 (June 2024): Bringing the Vision to Life:</h4><p>With the basics in place, it’s time to add some style! I design a beautiful interface within Ushahidi where users can easily manage their WhatsApp messages. It’s like decorating a house, making it welcoming and easy to use for everyone.</p><h4>Week 9–12 (July 2024): Putting It to the Test:</h4><p>Now that the project is taking shape, it’s time to see how well it works. I test every part of it to make sure everything runs smoothly. If I find any problems, I fix them right away. It’s like giving the project a final polish before it’s ready to go out into the world.</p><blockquote>This timeline shows how I move step by step through the project, from setting the environment to putting the finishing touches. Each week brings new challenges and accomplishments, helping me learn and grow along the way with the amazing Ushahidi team members.</blockquote><h3>My Thoughts and Discoveries:</h3><p>This project has been a rollercoaster of excitement and learning and I’m really excited for this amazing challenge journey because as always I’m a challenge lover. I’ve discovered how powerful technology can be when used for good. By connecting WhatsApp with Ushahidi, we’re giving people an easier way to ask for help and making sure responders can find them quickly when they need it most.</p><h4>🎯 My Testing Adventure:</h4><p>Testing was like playing detective. I had to make sure every button did what it was supposed to, every message got where it needed to go, and everything looked right. Sometimes things didn’t work, but that’s okay — it just meant there was more to learn and fix!</p><h4>Reflections on Testing WhatsApp Integration</h4><p>During this project, I took on the challenge of testing our WhatsApp integration feature using Cypress, a powerful testing framework, along with the Infobip API. <br>Let me walk you through the process and share my findings:</p><h4>Creating Basic Test Cases with the Infobip API:</h4><p>I crafted a series of test cases to ensure the reliability and functionality of our WhatsApp integration using Cypress and the Infobip API. Each test simulated real-world scenarios, such as sending messages to specific WhatsApp numbers or verifying successful delivery.</p><pre><br>// Cypress test case for Sending a WhatsApp Text Message via Infobip API<br><br>describe(&#39;Sending WhatsApp Text message via Infobip API&#39;, () =&gt; {<br>  it(&#39;Sends a WhatsApp text message&#39;, () =&gt; {<br><br>    // Define the message payload for sending a text message<br>    const textMessage = {<br>      from: &#39;your_whatsapp_number&#39;,<br>      to: &#39;recipient_whatsapp_number&#39;,<br>      messageId: &#39;unique_message_id&#39;,<br>      content: {<br>        text: &#39;Hello, this is a test message!&#39;<br>      }<br>    };<br><br>  // Send a POST request to Infobip API for sending a text message<br>    cy.request({<br>      method: &#39;POST&#39;,<br>      url: &#39;{your_api_base_url}/whatsapp/1/message/text&#39;,<br>      headers: {<br>        &#39;Content-Type&#39;: &#39;application/json&#39;,<br>        Authorization: {your_infobip_api_key}<br>      },<br>      body: textMessage<br>    }).then((response) =&gt; {<br><br>      // Assert that the response status is 200 (OK)<br>      expect(response.status).to.eq(200);<br><br>      // Log the response body<br>      cy.log(JSON.stringify(response.body));<br><br>       // Verify the status code is 200<br>       expect(response.status).to.eq(200);<br>    });<br>  });<br>});<br><br><br><br>// ---------------------<br><br><br>// 2nd Cypress test case for Sending a WhatsApp Document message via Infobip API<br><br>describe(&#39;Sending WhatsApp Document message via Infobip API&#39;, () =&gt; {<br>    it(&#39;Sends a WhatsApp document message&#39;, () =&gt; {<br>        <br>        // Define message data<br>        const messageData = {<br>            from: &#39;your_whatsapp_number&#39;,<br>            to: &#39;recipient_whatsapp_number&#39;,<br>            messageId: &#39;unique_message_id&#39;,<br>            content: {<br>                mediaUrl: &#39;your_document_url&#39;<br>            }<br>        };<br><br>        // Send request using Cypress<br>        cy.request({<br>            method: &#39;POST&#39;,<br>            url: &#39;{your_api_base_url}/whatsapp/1/message/document&#39;,<br>            headers: {<br>              &#39;Content-Type&#39;: &#39;application/json&#39;,<br>              Authorization: {your_infobip_api_key}<br>            },<br>            body: messageData<br><br>        }).then((response) =&gt; {<br>            // Verify the response status is 200 (OK)<br>            expect(response.status).to.eq(200);<br><br>            // Log the response body<br>            cy.log(JSON.stringify(response.body));<br>        });<br>    });<br>});</pre><h4>This Test Case Execution:</h4><p>Executing these test cases allowed me to assess the effectiveness of our WhatsApp integration. By sending messages using the Infobip API and monitoring the responses, I verified that everything worked as intended. These tests provided valuable insights into the reliability and functionality of our system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j9LUR8MIqkCwH3mJE0v0hw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*YW9qTsvECUrIPvjDburItA.png" /></figure><h4>Results and Insights:</h4><p>All my test cases passed successfully! This means my exploration of the WhatsApp business API integration feature is robust and reliable. However, I didn’t just stop at success; I also learned from failures with the pure mock data. When a test case didn’t pass, I investigated the issue, identified the root cause, and made the necessary adjustments to improve the system.</p><h4>Future Enhancements:</h4><p>While my starting integration is in good shape, there’s always room for improvement. In the future, I could explore adding features like multimedia message support or real-time analytics. Additionally, gathering user feedback could provide valuable insights for refining our system and making it even more user-friendly.</p><p>Testing the WhatsApp integration using Cypress and the Infobip API was a rewarding experience of mine that confirmed the strength of our project. By ensuring its reliability, we are one step closer to empowering marginalized communities and revolutionizing crisis response efforts.</p><h3>What’s Coming Up Next🤔?</h3><p>Even though we haven’t started our project yet, there’s a lot we can do to make it even cooler. We could add neat stuff like sending pictures, audios, videos, documents, or even booking meetings. And how about seeing messages on a map? That would be super cool! Also, listening to what people using our system have to say can help us make it even more helpful. So, let’s stay open-minded and get ready to make our Ushahidi project even better!</p><h3>Conclusion:</h3><p>Now, looking back on my exploration journey, I’m amazed at how much technology can help people. By putting together WhatsApp and Ushahidi, we’re not just making a tool—we&#39;re creating a lifeline for those who need it most. I’m thankful for the chance to be a part of this project as a contributor and excited to see what happens next.</p><blockquote><em>In sharing my story in this blog post, I hope to inspire others to use their skills to make the world better. Together, we can use technology to solve real problems and change lives for the better. Let’s work together to make a difference in the world 🌐.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f9305708072c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Test the Ushahidi Platform for Accessibility]]></title>
            <link>https://medium.com/@ilmasalsabil1006/how-to-test-the-ushahidi-platform-for-accessibility-94f4e2238c81?source=rss-bd636ef57a4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/94f4e2238c81</guid>
            <dc:creator><![CDATA[Ilmasalsabil]]></dc:creator>
            <pubDate>Tue, 19 Mar 2024 15:26:50 GMT</pubDate>
            <atom:updated>2024-05-19T06:32:44.564Z</atom:updated>
            <content:encoded><![CDATA[<p>Ensuring that the <a href="https://www.ushahidi.com/"><strong>Ushahidi</strong></a><strong> platform</strong> is accessible to all users, including those with disabilities, is crucial for providing an inclusive user experience. In this guide, I will provide you with step-by-step instructions on how to test the Ushahidi platform for accessibility using various tools and techniques.</p><p><strong>1. Understanding Accessibility Guidelines:</strong></p><p>Before diving into testing, it’s essential to familiarize yourself with the <a href="https://www.w3.org/TR/WCAG21/"><strong><em>Web Content Accessibility Guidelines (WCAG)</em></strong></a>. WCAG provides a set of standards for making web content accessible to people with disabilities. Understanding these guidelines will help you identify and address accessibility issues effectively.</p><p><strong>2. Using Accessibility Testing Tools:</strong></p><p>There are several tools available to help automate the process of accessibility testing. We’ll use the following tools in our testing process:</p><p>• <strong>axe CLI:</strong> An open-source command-line tool for testing web accessibility.<br>• <strong>axe Browser Extension:</strong> A browser extension that integrates axe accessibility testing directly into your browser.<br>• <strong>Cypress:</strong> A testing framework that allows you to write and run automated tests for web applications.</p><p><strong>3. Setting Up Testing Environment:</strong></p><p>• <a href="https://www.npmjs.com/package/@axe-core/cli"><strong>Install axe CLI globally</strong></a> on your machine using a cmd : npm install @axe-core/cli -g<br>• Install the <a href="https://www.deque.com/axe/browser-extensions/"><strong>axe browser extension</strong></a> in your preferred web browser.<br>• Set up Cypress for automated testing by following the installation instructions on the <a href="https://docs.cypress.io/guides/getting-started/installing-cypress"><strong>Cypress website</strong></a>.</p><p><strong>4. Conducting Accessibility Tests:</strong></p><ul><li>Start by navigating to different pages and components of the Ushahidi platform.<br>• Use the axe browser extension to run accessibility tests on each page. The extension will highlight any accessibility issues found.</li></ul><h3>Accessibility Issue: Map Page Button component lacks discernible text for screen readers</h3><p><a href="https://github.com/ushahidi/platform/issues/4825"><strong>Issue:</strong></a><strong> <br></strong> <a href="https://github.com/ushahidi/platform/issues/4825">https://github.com/ushahidi/platform/issues/4825</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YkUDyLNJ9w2gIgIGgVZN1w.png" /></figure><p>After finding and resolving the issue:</p><p><strong>Solving Issue </strong><a href="https://github.com/ushahidi/platform-client-mzima/pull/882"><strong>:</strong></a><strong> <br></strong> <a href="https://github.com/ushahidi/platform-client-mzima/pull/882">https://github.com/ushahidi/platform-client-mzima/pull/882/</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*16dkMoik-fWGSn6ICy0VFg.png" /></figure><h3>Integrating Axe cli to evaluate accessibility in an existing codebase</h3><p><a href="https://github.com/ushahidi/platform-client-mzima/pull/893"><strong>https://github.com/ushahidi/platform-client-mzima/pull/893</strong></a></p><ul><li>Utilize axe CLI to run batch tests on multiple pages or components of the platform. This can be done by writing custom scripts or integrating axe CLI into your CI/CD pipeline.</li><li>Using Axe cli tool to find the accessibility violation critical issues:</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LMkfPRiiH7YcPgZz5g-h9g.png" /></figure><ul><li>After Resolving the accessibility violation issue, you found 0 violation -</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uHo6y_yqFp3T76Mp2x4FUQ.png" /></figure><p>it helps us a lot to find the issue and get it fixed with ease</p><p>• Also, we can write Cypress tests to automate accessibility testing of interactive features and workflows on the platform. Ensure that these tests cover common user interactions and scenarios.</p><p><strong>5. Analyzing Test Results:</strong></p><p>• Review the accessibility issues identified by the testing tools. Pay attention to the severity of each issue and prioritize them based on their impact on users. <br>• Document the findings, including screenshots and descriptions of each issue.<br> • Categorize the issues based on WCAG guidelines and provide recommendations for remediation.</p><p><strong>6. Iterating and Improving:</strong></p><p>• Test the platform again after implementing fixes to ensure that the issues have been resolved.<br>• Continuously monitor and test the platform for accessibility as new features are developed and deployed.</p><h3><strong>Additional Insights on the Problem-Solving Approach:</strong></h3><p><strong>a. Identification Process:</strong></p><ul><li>Each accessibility issue mentioned in the blog post was first identified through a combination of manual inspection and automated testing using tools such as axe CLI, Cypress, and the axe browser extension. This ensured comprehensive coverage of potential accessibility barriers.</li></ul><p><strong>b. Resolution Strategies:</strong></p><ul><li>Upon identifying an accessibility violation, I prioritized addressing issues based on their impact and severity as per WCAG guidelines. For instance, critical issues affecting essential functionality or content were given immediate attention.</li><li>To resolve these issues, I followed best practices recommended by accessibility experts, which often involved modifying code elements, adjusting design components, or enhancing interactive features to meet accessibility standards.</li><li>Collaboration with developers and designers played a crucial role in implementing effective solutions, ensuring that accessibility considerations were integrated seamlessly into the development process.</li></ul><h3><strong>Best Practices and Recommendations:</strong></h3><ul><li>Throughout the resolution process, I adhered to industry best practices and standards, such as those outlined in the Web Content Accessibility Guidelines (WCAG). This ensured that the solutions implemented were not only effective but also aligned with recognized accessibility principles.</li><li>I documented the remediation process, providing clear explanations of the issues encountered, the steps taken to address them, and references to relevant accessibility guidelines. This documentation serves as a valuable resource for future reference and knowledge sharing within the team.</li></ul><h3><strong>Continuous Improvement:</strong></h3><ul><li>Recognizing that accessibility is an ongoing commitment, I emphasized the importance of continuous monitoring and testing. Regular audits and evaluations are essential to maintain compliance and address any emerging accessibility challenges as the platform evolves.</li><li>By adopting these strategies and practices, I believe we can ensure that the Ushahidi platform remains accessible to all users, reflecting our commitment to inclusivity and equality.</li></ul><h3><strong><em>Conclusion:</em></strong></h3><p>By following these steps and leveraging the power of axe CLI, axe browser extension, and Cypress, you can effectively test the Ushahidi platform or any other software for accessibility. Ensuring accessibility not only enhances the user experience but also reflects a commitment to inclusivity and equality for all users.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=94f4e2238c81" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>