<?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 Biljana Jelić on Medium]]></title>
        <description><![CDATA[Stories by Biljana Jelić on Medium]]></description>
        <link>https://medium.com/@biljana989?source=rss-164142e6624------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*g6e4AlVTQb3DIX9LaQtw3g.jpeg</url>
            <title>Stories by Biljana Jelić on Medium</title>
            <link>https://medium.com/@biljana989?source=rss-164142e6624------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 05:08:34 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@biljana989/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[SQL reserved keywords]]></title>
            <link>https://medium.com/towards-data-engineering/sql-reserved-keywords-925c0a4eee72?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/925c0a4eee72</guid>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[sql-server]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 11:22:06 GMT</pubDate>
            <atom:updated>2026-10-30T11:33:45.352Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r6CmNBtPg1f711VnyVdtYw.jpeg" /></figure><h4>Can you use them as column names?</h4><p>The other day, a friend of mine made a bank transfer to a company that, for the purpose of this article, I’ll call “<em>Transfer</em>”. (This is a made-up name, but the logic is similar to the real company name.) The payment was immediately flagged, and the bank reached out to ask: Is this a real payment? Who is “<em>Transfer</em>”, and why are you transferring money to “<em>Transfer</em>”?</p><p>We ended up talking about how that probably isn’t the best name for a company. The bank’s security system was obviously flagging the payment because the name sounded suspicious, even though it was a legitimate transaction.</p><p>That conversation made me think about SQL reserved words. Sometimes, I would name a column or a table based on what makes sense in my context, forgetting that the same word might already have a special meaning in SQL Server, which can then cause unexpected issues. The most recent example is the word DESC, which is in my mind a perfectly reasonable abbreviation for <em>description</em> (while I create a temp table to store some codes and their descriptions), but SQL doesn’t read it in the same way, because in SQL DESC has a completely different meaning. And that’s where things get a little confusing.</p><h4>When a column name becomes a problem</h4><p>In SQL Server, some words have a special meaning because they’re part of the SQL language itself. These are called keywords, and some of them are reserved keywords.</p><p>Take DESC.It’s used in an ORDER BY clause to sort results in descending order, from highest to lowest:</p><pre>SELECT *<br>FROM #TempTable<br>ORDER BY 1 DESC;</pre><p>Now say you’ve created a temporary table with a column called DESC, intending it to hold the description of each code, and you run a simple SELECT:</p><pre>SELECT DESC<br>FROM #TempTable;</pre><p>Instead of results, SQL Server gives you an error:</p><p><em>Incorrect syntax near the keyword &#39;DESC&#39;.</em></p><h4>The simple fix</h4><p>The good news is that even though DESC is a reserved keyword, you can still use it as a column name. The workaround is very simple: put square brackets around it.</p><pre>SELECT [DESC]<br>FROM #TempTable;</pre><p>The brackets tell SQL Server to treat DESC as an identifier (in this case, a column name) rather than as part of the SQL syntax.</p><p>This isn’t limited to column names. It works for table names, temp tables and aliases too. For example, AS [Order] will work, while a plain AS Order will give you an error. Brackets also help with names that contain spaces or special characters, and with reserved words you might come across in existing databases</p><h4>Brackets as a habit</h4><p>Most ordinary column names don’t need brackets, but it’s good practice to use them anyway. When thinking about good SQL practices, it’s worth looking at how SQL Server itself handles things. If you generate scripts with the SSMS wizard, you’ll notice it automatically puts brackets around every column name. That way, there’s no confusion about how SQL should interpret each word.</p><p>Keep in mind that square brackets are specific to SQL Server. Other databases have their own way of doing the same thing. PostgreSQL uses double quotes &quot;DESC&quot;, and MySQL uses backticks `DESC`. So if you move your code between systems, this is one of the things you might need to change.</p><h4>A quick way to spot reserved words</h4><p>Another way to recognise a reserved word is by the colour of the text in SSMS. Keywords are highlighted in blue. So if you hit an error and your column name is showing up in blue, you’re probably using a reserved word. Add brackets, and it will work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/622/1*90V-_SkPRAmLy-1PfYmxwA.png" /></figure><h4>List of reserved words</h4><p>To give you a better idea of which words to watch out for, here are some of the most common column names that people forget are reserved words:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aIhoEhjOxzgE4E59aSoodQ.png" /><figcaption>Common SQL reserved words</figcaption></figure><p>The full list is much bigger, and you can find it on <a href="https://learn.microsoft.com/en-us/sql/t-sql/language-elements/reserved-keywords-transact-sql?view=sql-server-ver17">Microsoft’s reserved keywords page</a>.</p><p>There is one more thing worth knowing. Besides the reserved words, Microsoft also lists words that aren’t reserved yet but could become reserved in future versions of SQL Server. If you use one of them as a column name today, it might break after an upgrade. They don’t cause errors right now, so you will not notice them, but Microsoft recommends avoiding them. You will find them on the same Microsoft page, under the <strong>future reserved keywords section.</strong></p><h4>End</h4><p>I love a good full-circle moment. Looking back, my friend’s bank transfer and my DESC column had more in common than I initially thought. Both were perfectly fine, but both looked suspicious to a system that has its own idea of what that word means. The bank needed a phone call, and SQL Server just needs a pair of brackets. And if we take this beyond SQL and into everyday life, there’s an interesting parallel. We often assume that others interpret words, but the same word, action or situation can mean something completely different depending on who’s interpreting it. Sometimes, it’s worth looking beyond our own assumptions before jumping to conclusions.</p><p>Thank you for reading.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=925c0a4eee72" width="1" height="1" alt=""><hr><p><a href="https://medium.com/towards-data-engineering/sql-reserved-keywords-925c0a4eee72">SQL reserved keywords</a> was originally published in <a href="https://medium.com/towards-data-engineering">Towards Data Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Trying to keep with AI]]></title>
            <link>https://biljana989.medium.com/trying-to-keep-up-in-the-age-of-ai-a4085a3c8532?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/a4085a3c8532</guid>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Wed, 30 Sep 2026 19:04:19 GMT</pubDate>
            <atom:updated>2026-10-24T18:22:15.328Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s9ugeGPnon4vG3XEJGZplA.jpeg" /></figure><p>Last day of September, everyone is fully back, and summer is fading and becoming a memory. This was the first summer when I noticed that quite a few people close to me didn’t really take a proper one or two weeks off. And those who did, me included, seemed to come back with a slightly unexpected feeling of being out of the loop. It made me wonder whether this feeling is something relatively new as technology keeps accelerating, and what it actually means to disconnect when there is always something new happening.</p><p>Being offline somehow created a backlog of things to catch up on. While we were away, the tech world didn’t take any days off.</p><p>Meta launched a new AI coding agent, SpaceX acquired Cursor, and Lovable raised $400 million and expanded into MCP and almost 100 app connectors. Anthropic talked about increasingly capable models and AI safety, including incidents involving Claude models and real computer systems. And there was a whole list of other things that apparently you should know about.</p><p>Today, saying that you haven’t checked any news for a few weeks can almost sound like a negative personal choice, as if you have neglected some kind of responsibility. Maybe that’s because information is no longer something we simply receive without thinking about how it affects our lives. It has become something we actively consume every day, and with that comes a certain responsibility to understand what is happening around us, rather than blindly accepting every new change simply because it is new. There is also a responsibility that comes from the fact that we actively choose what we use and what we support. So yes, we should stay informed, at least to some extent, and pay attention to what is happening around us. But how much is enough? If the world never stops updating, being fully up to date isn’t really a realistic state. That might sound a little disturbing, but the awareness that we simply cannot be faster than technological progress could actually give us some peace of mind, rather than becoming another source of pressure to keep up. It is pretty obvious that we can’t outpace technology, but our minds have a funny way of dwelling on things that we already know to be true.</p><p>There is probably no single way to experience this constant acceleration. How we experience it, and how much it affects us, will be different for everyone. It goes beyond the usual work-life balance conversation and starts to touch on how we relate to technology, how we think about progress, and how we see ourselves in a world that keeps moving forward. I don’t want to turn this into a cheesy “how to feel better” article. Instead, I want to mention a couple of concepts that I find interesting when thinking about productivity and the pace we are living in, and that might give you something to reflect on.</p><h3><strong>Parkinson’s Law</strong></h3><p>The first one is Parkinson’s Law: <strong>“W<em>ork expands so as to fill the time available for its completion.”</em></strong></p><p>It is interesting to think about this one in the context of AI, because AI can improve our productivity, but that doesn’t necessarily mean the amount of work we feel we need to do will decrease. In fact, it can very easily go in the opposite direction.</p><p>The faster technology makes things, the more things we can do. The more things we can do, the more we expect ourselves to do. And the more we expect ourselves to do, the more things there are to keep up with. So instead of simply getting more free time because we can get more things done in a shorter amount of time, we might end up with higher expectations and more things on our plate.</p><p>That is one of the paradoxes of using AI in our daily lives:<strong> the technology gets faster, but somehow our feeling of having enough time doesn’t necessarily follow.</strong></p><h3><strong>“Frenetic standstill”</strong></h3><p>The second concept is “<strong>frenetic standstill</strong>”. It describes the feeling of doing more and moving faster, while somehow not feeling like you are making meaningful progress. It captures something very specific about the feeling of being more productive than ever and still somehow feeling behind.</p><p>There is something contradictory about the way we experience technology, and we are starting to see it even more clearly with AI. Its initial promise was to make our lives easier, save us time and help us get more done, but in many ways it can also do the opposite.</p><p>It will be really interesting to see what psychology and research tell us about what happens when we start living and working alongside increasingly capable AI systems in the future. Maybe there is already a lot of research on this, but I haven’t caught up on that yet. 😊</p><p>If you’ve come across anything interesting on this topic, please share it in the comments. I’d love to hear about it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a4085a3c8532" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to learn SQL in the age of AI?]]></title>
            <link>https://generativeai.pub/how-to-learn-sql-in-the-age-of-ai-8b7fd59b5e70?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/8b7fd59b5e70</guid>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Tue, 09 Jun 2026 08:20:37 GMT</pubDate>
            <atom:updated>2026-06-16T08:34:05.916Z</atom:updated>
            <content:encoded><![CDATA[<h4>If AI is now assisting with query writing, what should students focus on learning?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bHOBu60zw_3zwFyhZjMAPw.png" /></figure><p>Database courses were always among my favourite subjects at university. I always thought they were one of those subjects that made me feel like I was learning something I could actually use in the real world. There was something satisfying about SQL. Unlike some other programming languages that felt abstract and intimidating at first, SQL felt practical. I also had a great professor, who made a real difference, and more than once, I caught myself imagining what it would be like to stand at the front of the classroom one day and teach the same subject.</p><p>Our labs and assignments required us to build queries ourselves. We memorised syntax because we had to. We learned the differences between INNER JOINs, LEFT JOINs, and RIGHT JOINs and the difference between procedures and functions. We practised writing queries by hand, and exams often tested whether you could remember the correct syntax under pressure, and success was typically measured by whether your query produced the expected result. Back then, learning SQL was probably a different experience from what it is today.</p><p>When you got stuck, you had to figure things out yourself. Sometimes that meant reading documentation. Sometimes it meant searching through Stack Overflow for an answer posted by someone who had exactly the same problem six years earlier. Sometimes it meant staring at your query for an hour before finally noticing the mistake. There was no ChatGPT to help you spot the issue in a second. It might sound time-consuming, but a surprisingly large amount of learning happened during those moments. We learned because we got stuck.</p><p>Today, I often wonder what that learning journey looks like for someone opening SQL for the first time. That’s why I’m writing this, and I don’t have the answers, nor do I know exactly how SQL is learned today, but I believe the traditional model of learning is becoming outdated. With AI tools capable of generating syntactically correct SQL in seconds, memorisation has less value than it once did. AI is driving a similar shift in many other areas, making memorisation-based learning less important than it once was. It’s no longer realistic to expect students to write complex, multi-join queries entirely from scratch without assistance. In some ways, that’s a positive shift. Less time spent debugging minor syntax errors means more time can be spent building meaningful projects and solving real problems. In theory.</p><blockquote><strong>But if AI is now assisting with query writing, what should students focus on learning?</strong></blockquote><p>Learning SQL has traditionally been about practice and repetition; the more queries you write, the more patterns you start to see, and the more confident you become. It’s all about practice.</p><p>I do believe that syntax still matters, but it is no longer the centrepiece. What matters more is the ability to think in terms of data, use tools effectively, and critically evaluate results. If AI is now assisting with query writing, students need to focus more on interpreting results, validating correctness, and refining logic. They need to understand how data relates across tables, how queries transform that data, and whether the output actually answers the original question. And understanding core concepts like filtering, grouping, and joining data is essential. Without that foundation, it’s difficult to approach a problem or anticipate what the result should look like.</p><p>It’s also important to understand what AI tools like ChatGPT actually do. They are not “thinking”, they generate queries based on patterns learned from large amounts of data. The results can look convincing while still being incorrect. That is where education should increasingly focus: recognising when something is wrong and understanding why.</p><h3><strong>How I Would Teach SQL Today</strong></h3><p>So, if I were teaching SQL in schools today, I wouldn’t treat it as a traditional programming subject built around memorising syntax and writing queries from scratch. With AI tools now available, the focus should shift towards understanding data and using SQL as a practical tool. Purely abstract learning is becoming less effective, especially when students can rely on AI to provide immediate support when they get stuck. With AI now helping to explain and break down complex ideas, learning should feel more practical and purposeful. I would start with real, concrete problems that students actually care about and can relate to. Instead of learning SQL commands in isolation or working with made-up examples made for textbooks, they could practise using real data available online and use data to answer questions, investigate patterns, and explore topics that interest them, like sports statistics, music trends, movie ratings, public transport data and similar. SQL would simply be the tool used to find those answers. This makes learning more engaging and shows how SQL is used in the real world.</p><p>AI tools would be introduced early, but positioned carefully as assistants rather than replacements for thinking. Students could use AI to generate queries, but they shouldn’t stop there. The important part of the exercise would be interpreting what the AI produced, explaining each part of the query, checking whether it actually answers the question, and adjusting it when it doesn’t. In practice, this is quite close to how SQL is already used in real environments: not written perfectly from memory, but iterated on, reviewed, and corrected.</p><p>Instead of covering everything in SQL, I would prioritise building a strong mental model of relational data. Students should understand what tables represent, how rows and columns map to real-world entities, and how relationships between tables work. Without that foundation, AI-generated queries remain abstract and difficult to understand. The model should be clear, so that concepts like joins, filtering, grouping, and sorting become easier to grasp.</p><p>I would treat AI mistakes as part of the learning process. AI-generated queries are often slightly wrong, using incorrect columns, misunderstanding relationships, or producing inefficient logic. Rather than avoiding that, I would build lessons around it. Students would diagnose why a query fails or produces incorrect results and then fix it. This develops a more realistic and transferable skill set, as well as critical thinking. We still need to keep exercising our thinking skills and mental muscles, even when tools can do part of the work for us.</p><p>Following that, tests would focus on how well students can interpret, critique, and adapt existing queries. Being able to explain what a query does, identify issues, and modify it to answer a new question is a stronger indicator of understanding in an AI-assisted environment.</p><p>SQL should also be seen in a broader context. It is just one tool in a larger data ecosystem. With modern tools available, it’s important to recognise when SQL is the right choice and when other approaches might be more effective. That awareness reflects how data work actually happens today, where tools are chosen based on the problem rather than taught in isolation.</p><p>The goal is no longer to produce students who can write complex SQL without help, but those who can think clearly about data, use AI effectively, and critically evaluate outputs. That combination is far more aligned with how technology is used today.</p><p>In many ways, SQL education is reaching a turning point, not because SQL is less important, but because the skills around it are changing.</p><p>If you’ve seen real examples, whether a student, intern, or junior developer who has learned recently, I’d love to hear how they studied. What worked, what didn’t, and how AI has changed the process for them?</p><p>Thank you for reading.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*B7o5L3Jtc-w-Z0TO.png" /></figure><p>This story is published on <a href="https://generativeai.pub/">Generative AI</a>. Connect with us on <a href="https://www.linkedin.com/company/generative-ai-publication">LinkedIn</a> and follow <a href="https://www.zeniteq.com/">Zeniteq</a> to stay in the loop with the latest AI stories.</p><p>Subscribe to our <a href="https://www.generativeaipub.com/">newsletter</a> and <a href="https://www.youtube.com/@generativeaipub">YouTube</a> channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*2oQmlpEMdHjLPxyw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8b7fd59b5e70" width="1" height="1" alt=""><hr><p><a href="https://generativeai.pub/how-to-learn-sql-in-the-age-of-ai-8b7fd59b5e70">How to learn SQL in the age of AI?</a> was originally published in <a href="https://generativeai.pub/">Generative AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Silent SQL Trap: LEFT JOIN WHERE vs ON]]></title>
            <link>https://biljana989.medium.com/the-silent-sql-bug-left-join-where-vs-on-trap-d70a15ec0a3a?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/d70a15ec0a3a</guid>
            <category><![CDATA[sqlbug]]></category>
            <category><![CDATA[sql-left-join]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Mon, 27 Apr 2026 09:46:56 GMT</pubDate>
            <atom:updated>2026-05-04T10:11:56.511Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qgqQ4vASl8dwq7Tsb6f0ng.png" /></figure><p>Let’s talk about another SQL behaviour that looks harmless until it quietly changes your results. You write what feels like a simple query, it runs fast, has no errors, and the result looks clean. Then someone notices that the numbers don’t quite match, data is missing, and you spend hours debugging to find <strong>one line</strong> that turned your safe and simple query into a little headache.</p><p>Let’s simplify with an example. Say we want all Orders, but only successful Payments if they exist, so we write the following SQL:</p><p><em>P.S. I’ll write Orders and Payments in uppercase to emphasise that they are table names in this example.</em></p><pre>SELECT o.id, p.status<br>FROM orders o<br>LEFT JOIN payments p <br>  ON o.id = p.order_id<br>WHERE p.status = &#39;SUCCESS&#39;;</pre><p>So the SQL logic here is simple but strict. It first takes all Orders and tries to match them with Payments using LEFT JOIN. Every Order gets a row, and unmatched Orders fill Payments columns with NULLs, which is standard and expected LEFT JOIN behaviour. But then WHERE runs on this complete result to check if p.status = &#39;SUCCESS&#39; . Since NULL doesn&#39;t equal &#39;SUCCESS&#39;, every row with NULL status disappears. So in this instance, the WHERE clause removes all rows where p.statusis NULL, excluding all Orders without Payments and those with non-successful Payments from the query. So even though we started with a LEFT JOIN, we end up only seeing Orders that have successful Payments. <strong>The LEFT JOIN quietly behaves like an INNER JOIN</strong>, with no warning at all, which is probably not what you intended.</p><p>Correct version would be:</p><pre>SELECT o.id, p.status<br>FROM orders o<br>LEFT JOIN payments p <br>  ON o.id = p.order_id <br> AND p.status = &#39;SUCCESS&#39;;</pre><p>Now this version works correctly. The ON clause controls the join, and only Payments whereo.id = p.order_id AND p.status = &#39;SUCCESS&#39; get attached to Orders. All Orders remain in the result set, with NULL values appearing in the Payments columns for unmatched records, as expected from the LEFT JOIN. WHERE operates later on the complete result and cannot eliminate left table rows.</p><blockquote><strong>The key difference lies in SQL’s execution order, and when the condition is applied. </strong>The ON clause controls how rows are joined, while WHERE filters the final result after everything is already combined.</blockquote><p>Before writing your query, try to ask yourself this simple question: “Does this condition decide which rows join together (ON), or does it filter the final results (WHERE)?” This one mental check can be crucial when queries grow. A single condition in the wrong place can quietly remove half your data.</p><p>Like other SQL traps I’ve covered, everything comes down to one line of code. If you move it from WHERE to ON, and missing data remains visible instead of being filtered out.</p><h3>INNER JOIN is different</h3><p>This wouldn’t be so noticeable in the INNER JOIN because INNER JOIN already removes non-matching rows, WHERE filters after and ON filters during.</p><pre>SELECT o.id, p.status<br>FROM orders o<br>INNER JOIN payments p <br>  ON o.id = p.order_id<br>WHERE p.status = &#39;SUCCESS&#39;;<br><br>SELECT o.id, p.status<br>FROM orders o<br>INNER JOIN payments p <br>  ON o.id = p.order_id <br> AND p.status = &#39;SUCCESS&#39;;</pre><p>There is still a difference, but more subtle as ON defines matching logic and WHERE defines final filtering. One filter results, the other defines the relationship. Both give the same results here. But ON can be faster because it filters while joining.</p><h4><strong>SQL does this order:</strong></h4><ol><li>Join tables together</li><li>Filter with WHERE</li><li>Pick columns with SELECT</li></ol><p>LEFT JOIN makes rows with NULLs for missing matches. Then WHERE sees those NULLs and removes them when you filter on the right table.</p><p>If you want to catch this trap early, you can<strong> do a simple test firs</strong>t. Run the query without the WHERE clause and check the number of rows:</p><pre>SELECT COUNT(*) as all_rows<br>FROM orders o<br>LEFT JOIN payments p ON o.id = p.order_id;</pre><p>Then add your condition. If the number drops below your total number of Orders, your LEFT JOIN is no longer behaving the way you expect.</p><h3><strong>Why are we making this mistake?</strong></h3><p>LEFT JOIN + WHERE is a very misunderstood pattern because a lot of people assume:</p><blockquote><em>LEFT JOIN = keep everything from left table</em></blockquote><p>But the real rule is:</p><blockquote><em>LEFT JOIN keeps everything </em><strong><em>only until another condition removes it</em></strong></blockquote><p>When we write LEFT JOIN ... WHERE right_table.column = valuewe expect the left table to stay preserved. But SQL follows strict logic, and it evaluates WHERE on the final joined result. NULL values from unmatched rows fail the condition, so those rows disappear entirely.</p><h4>Simple rule to remember</h4><p>Before adding any condition, ask yourself: “If the Payment table was empty, should I still see orders?”</p><ul><li>If the answer is Yes → put conditions in ON</li><li>the answer is No→ put conditions in WHERE</li></ul><p>That’s another simple trick we should practice.</p><p>Like many SQL bugs, this one comes down to a single line of code. Moving a condition from WHERE to ON can completely change your results, even though the query still looks correct. SQL does not warn you when your logic is wrong. It just returns data, so before running any query, try to think through what you actually expect to get back. And sometimes, that data looks perfectly fine until you realise something is missing.</p><p>Thank you for reading!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d70a15ec0a3a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[3 small SQL tasks I let AI handle]]></title>
            <link>https://biljana989.medium.com/3-tiny-sql-tasks-i-let-ai-handle-a405340bb5b0?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/a405340bb5b0</guid>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Wed, 18 Mar 2026 12:27:56 GMT</pubDate>
            <atom:updated>2026-03-19T12:16:29.711Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IJih4qAb4ZyP-56uwUnsQg.png" /></figure><p>AI has become a great help in assisting us with all kinds of tasks, and programming is no exception. Writing queries, suggesting better ways to structure them, or simply reminding us of the syntax we have forgotten. But that’s not what I want to write about because I know there are still people who prefer not to rely on AI to write their queries and would rather stick to the old-school way as much as possible. And I get that. But there are areas where it’s incredibly useful, and that’s what I want to focus on.</p><p>For me, the biggest benefit of AI is that there is finally something to handle the simple, boring, repetitive tasks that none of us enjoys doing. Things like spreadsheets, presentations, Excel cleanup, formatting, and similar chores. I don’t know anyone who likes that part of the job, and I’m sure neither do you. It’s time consuming, it is easy to mess up, and it&#39;s simply not worth our time.</p><p>So thank you, AI, for helping with all those little manual tasks that are easy but drain our energy, and after years of doing them, you just want to find a faster way.</p><blockquote>Those are the moments when I truly appreciate AI, when it does all these repetitive, boring tasks for me in seconds.</blockquote><p><strong>Here are my top 3 favourite tasks I let AI handle for me:</strong></p><h4><strong>1. </strong>Converting raw data into SQL ready format</h4><p>There are so many situations where you receive a raw list of values that you somehow need to get into the database. Maybe you want to use them in a WHERE ... IN (...) filter to compare the list against your existing data, or you need to turn them into proper INSERT statements. And of course, the list almost never arrives in a nice SQL ready format. It comes as a vertical list in Notepad, a column in Excel, a copied table from an email, or a random dump from a ticket description. But to simplify the example, let’s imagine the list like this:</p><pre>1023<br>1044<br>1088<br>1201</pre><p>Instead of manually adding commas, quotes, and brackets, I just paste whatever I have into AI and describe what I need. For example: “Here is a list of IDs, convert it into a single‑line SQL IN clause” or “Turn this list of codes into INSERT values for table X (column code).” In a second, I get exactly what I need, already formatted, like:</p><p>WHERE id IN (&#39;1023&#39;,&#39;1044&#39;,&#39;1088&#39;,&#39;1201&#39;)</p><p>I used to spend so much time manually editing these lists, using SQL formatting functions and reminding myself of the syntax, doing Excel magic, or search‑and‑replace just to get them into the format I needed, but now it is a few seconds of copy-paste into AI. I use it all the time when I need a horizontal layout for IN clauses, CASE statements, or quick SELECT ... WHERE id IN (...) filters.</p><p>This can go even further with slightly smarter prompts like: “Here’s a multi‑column list; use only the first column to build an IN clause.” In that case, you don’t even have to clean your input first – AI will do the picking and formatting for you.</p><h4><strong>2. </strong>Generating repeated SQL patterns</h4><p>Another thing I love is using AI to generate the same SQL pattern many times, instead of copy‑pasting and editing it by hand. Any time I have a repetitive SQL structure that needs to be repeated over and over with just a few values changing, AI is perfect for it.</p><p>For example, say I have this list in Excel or Notepad:</p><pre>John, IT<br>Maria, HR<br>Alex, Finance</pre><p>Instead of manually writing three INSERT statements, I paste this into AI and say: “Generate SQL INSERT statements for a table employees(name, department) from this list.” AI will then produce something like:</p><pre>INSERT INTO employees (name, department) VALUES (‘John’, ‘IT’);<br>INSERT INTO employees (name, department) VALUES (‘Maria’, ‘HR’);<br>INSERT INTO employees (name, department) VALUES (‘Alex’, ‘Finance’);</pre><p>This works great for almost any repetitive pattern: generating INSERT statements from a CSV export, creating a bunch of GRANT SELECT ON table_x TO user_y; statements for a list of tables, building CASE expressions from a mapping table, or producing long lists of old_column AS new_column mappings.</p><p>What I like is that I stay in control of the logic and the structure. I define the pattern once in my prompt, and AI just applies it consistently across all rows. Super fast, super accurate.</p><h4>3. Data formatting</h4><p>Probably my favourite use case is formatting. Commas, apostrophes, spaces, capitalisation, indentation, and my personal enemy: quotes. Just like in any programming language, handling single and double quotes in SQL can turn into a nightmare. But with AI, it takes just one simple prompt, and your formatting is done.</p><p>Here are some examples of what AI does for me now:</p><ul><li>Reformatting a messy query so it’s easier to read</li><li>Converting a list of column names into a comma‑separated list for a SELECT</li><li>Changing naming conventions (e.g., from your_name to yourName)</li><li>Adding consistent table aliases everywhere</li><li>Cleaning up string literals and date formats so they match the expected syntax without me double-checking every one.</li><li>Creating printed SQL queries in a specific format, like turning a query into a documentation-ready version with UPPERCASE keywords, consistent spacing, and clear comments for every section.</li></ul><pre>-- Before <br>SELECT customer_id,name FROM customers WHERE id IN(1,2,3)<br><br>-- After (AI formatted)  <br>SELECT c.customer_id, c.name <br>FROM customers c <br>WHERE c.id IN (1, 2, 3);</pre><p>All these tasks are micro cleanups, but they add up fast. Instead of spending 15–20 minutes cleaning a query, I let AI handle it while I focus on the next steps.</p><h3>Why this actually matters</h3><p>None of these tasks is “hard” work, but they’re exactly the kind that eats our time and energy. By letting AI handle them, we move faster from raw input (emails, notes, CSVs) to usable SQL. We also cut down on typos from repetitive typing and keep our focus on the actual logic and data, not juggling quotes and commas.</p><p>I’m writing this as a reminder of these small wins where AI can be our best assistant. If AI handles the boring stuff, we can do more interesting and creative work, and we should definitely take advantage of it.</p><p>I’m curious to hear what you’ve completely outsourced to AI. Let me know in the comments!</p><p>Thank you for reading.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a405340bb5b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The NOT IN + NULL Trap: Why Your SQL Query Returns No Rows]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://biljana989.medium.com/the-not-in-null-trap-why-your-sql-query-returns-no-rows-bbcd9bad47aa?source=rss-164142e6624------2"><img src="https://cdn-images-1.medium.com/max/1101/1*o7E7RHEOjLLlELnaVrRJ8A.png" width="1101"></a></p><p class="medium-feed-snippet">Continuing in the same spirit as my previous article, &#x201C;Silent SQL Bugs,&#x201D; I wanted to keep the series going with another one from the same&#x2026;</p><p class="medium-feed-link"><a href="https://biljana989.medium.com/the-not-in-null-trap-why-your-sql-query-returns-no-rows-bbcd9bad47aa?source=rss-164142e6624------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://biljana989.medium.com/the-not-in-null-trap-why-your-sql-query-returns-no-rows-bbcd9bad47aa?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/bbcd9bad47aa</guid>
            <category><![CDATA[sqlbug]]></category>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[null]]></category>
            <category><![CDATA[microsoft-sqlserver]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Wed, 25 Feb 2026 12:26:00 GMT</pubDate>
            <atom:updated>2026-02-25T12:26:00.794Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Silent SQL Bug : When An Invalid Column Name Doesn’t Throw an error]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/learning-sql/the-silent-sql-bug-when-an-invalid-column-name-doesnt-throw-an-error-b5c9277bc68a?source=rss-164142e6624------2"><img src="https://cdn-images-1.medium.com/max/1156/1*bSChsP1EzpEOoJMi-ikTAQ.png" width="1156"></a></p><p class="medium-feed-snippet">TL;DR: SQL Server may silently convert an invalid column in a subquery into a correlated reference, meaning your query runs without errors&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/learning-sql/the-silent-sql-bug-when-an-invalid-column-name-doesnt-throw-an-error-b5c9277bc68a?source=rss-164142e6624------2">Continue reading on Learning SQL »</a></p></div>]]></description>
            <link>https://medium.com/learning-sql/the-silent-sql-bug-when-an-invalid-column-name-doesnt-throw-an-error-b5c9277bc68a?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/b5c9277bc68a</guid>
            <category><![CDATA[sql-server]]></category>
            <category><![CDATA[sql-silent-bug]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Wed, 21 Jan 2026 15:14:03 GMT</pubDate>
            <atom:updated>2026-01-21T15:14:03.017Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How to Automate PDF Imports into SQL]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/towards-data-engineering/how-to-automate-pdf-imports-into-sql-07a33d2764c8?source=rss-164142e6624------2"><img src="https://cdn-images-1.medium.com/max/1536/1*TUX98ncv8WEs8da7_rsEhA.png" width="1536"></a></p><p class="medium-feed-snippet">&#x2014; A Step by Step Guide</p><p class="medium-feed-link"><a href="https://medium.com/towards-data-engineering/how-to-automate-pdf-imports-into-sql-07a33d2764c8?source=rss-164142e6624------2">Continue reading on Towards Data Engineering »</a></p></div>]]></description>
            <link>https://medium.com/towards-data-engineering/how-to-automate-pdf-imports-into-sql-07a33d2764c8?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/07a33d2764c8</guid>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[powershell]]></category>
            <category><![CDATA[csv]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Wed, 17 Sep 2025 16:44:18 GMT</pubDate>
            <atom:updated>2025-09-18T08:50:55.996Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[I Missed the World’s Largest Hackathon, but It Gave Me the Push I Needed]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://generativeai.pub/i-missed-the-worlds-largest-hackathon-but-it-gave-me-the-push-i-needed-cad28d60b0f2?source=rss-164142e6624------2"><img src="https://cdn-images-1.medium.com/max/1536/1*UcmlXEg4Iqv6GQSSrdx5_w.png" width="1536"></a></p><p class="medium-feed-snippet">A few days ago, The World&#x2019;s Largest Hackathon organized by Bolt came to an end. A massive global event, the one where they were attempting&#x2026;</p><p class="medium-feed-link"><a href="https://generativeai.pub/i-missed-the-worlds-largest-hackathon-but-it-gave-me-the-push-i-needed-cad28d60b0f2?source=rss-164142e6624------2">Continue reading on Generative AI »</a></p></div>]]></description>
            <link>https://generativeai.pub/i-missed-the-worlds-largest-hackathon-but-it-gave-me-the-push-i-needed-cad28d60b0f2?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/cad28d60b0f2</guid>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[boltnew]]></category>
            <category><![CDATA[hachathon]]></category>
            <category><![CDATA[lovable]]></category>
            <category><![CDATA[vibe-code]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Fri, 11 Jul 2025 10:52:17 GMT</pubDate>
            <atom:updated>2025-07-11T10:52:17.449Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Vibe Coding: A Hobby Today, the Future of Development Tomorrow?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://pub.towardsai.net/vibe-coding-a-hobby-today-the-future-of-development-tomorrow-a554f93c27a5?source=rss-164142e6624------2"><img src="https://cdn-images-1.medium.com/max/1536/1*Vfklk9nDH1-B5T1tXKN00w.png" width="1536"></a></p><p class="medium-feed-snippet">Vibe coding has become my latest hobby, and I must admit, it&#x2019;s pretty addictive&#x200A;&#x2014;&#x200A;unlocking a whole new level of technical creativity for&#x2026;</p><p class="medium-feed-link"><a href="https://pub.towardsai.net/vibe-coding-a-hobby-today-the-future-of-development-tomorrow-a554f93c27a5?source=rss-164142e6624------2">Continue reading on Towards AI »</a></p></div>]]></description>
            <link>https://pub.towardsai.net/vibe-coding-a-hobby-today-the-future-of-development-tomorrow-a554f93c27a5?source=rss-164142e6624------2</link>
            <guid isPermaLink="false">https://medium.com/p/a554f93c27a5</guid>
            <category><![CDATA[lovable]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Biljana Jelić]]></dc:creator>
            <pubDate>Thu, 24 Apr 2025 22:02:28 GMT</pubDate>
            <atom:updated>2025-04-25T18:20:37.587Z</atom:updated>
        </item>
    </channel>
</rss>