<?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 konrad bachusz on Medium]]></title>
        <description><![CDATA[Stories by konrad bachusz on Medium]]></description>
        <link>https://medium.com/@konradbachusz?source=rss-6ed77e3d609------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*R8r47vUev53__pvPi4GKXA.jpeg</url>
            <title>Stories by konrad bachusz on Medium</title>
            <link>https://medium.com/@konradbachusz?source=rss-6ed77e3d609------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 04:59:27 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@konradbachusz/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[How to avoid being automated by AI in 2025]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/credera-engineering/how-to-avoid-being-automated-by-ai-in-2025-958be4c797c7?source=rss-6ed77e3d609------2"><img src="https://cdn-images-1.medium.com/max/2041/1*1bH9ePMVrKhr8mKuRTIaIA.png" width="2041"></a></p><p class="medium-feed-snippet">Introduction</p><p class="medium-feed-link"><a href="https://medium.com/credera-engineering/how-to-avoid-being-automated-by-ai-in-2025-958be4c797c7?source=rss-6ed77e3d609------2">Continue reading on Credera Engineering »</a></p></div>]]></description>
            <link>https://medium.com/credera-engineering/how-to-avoid-being-automated-by-ai-in-2025-958be4c797c7?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/958be4c797c7</guid>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Thu, 10 Jul 2025 08:49:50 GMT</pubDate>
            <atom:updated>2025-07-10T08:49:50.452Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Beyond expectations: How AI has defied predictions]]></title>
            <link>https://medium.com/credera-engineering/beyond-our-expectations-how-ai-defied-predictions-7aa1c0d29aca?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/7aa1c0d29aca</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Thu, 10 Jul 2025 08:27:19 GMT</pubDate>
            <atom:updated>2025-07-23T10:38:09.416Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KGYHHk2dqfnpKE-II5RUng.png" /></figure><h3>Introduction</h3><p>For years, the promise of Artificial Intelligence (AI) has captivated our imaginations, often painting a picture of a distant, sci-fi future. Yet, as we stand here in July 2025, it’s clear that the trajectory of AI development has not only accelerated but has also taken us down paths few truly predicted.</p><p>This article dives into the unexpected capabilities of AI in 2025, exploring how the very areas we once thought were exclusively human domains have been fundamentally transformed, challenging our previous assumptions and proving that the future of AI is unfolding.</p><h3>What’s surprising about AI capabilities in 2025?</h3><p>The capability of different AI systems has been changing throughout the years. For this reason, the scope of this section is limited to what has been achieved up until June 2025. An interesting observation that I made is that the emergence of Deep Learning and the introduction of the Transformer neural network architecture in 2016 has taken the world by surprise. The book that greatly inspired me to learn more about AI was a book called <a href="https://amzn.to/4484R9M">“Life 3.0: Being Human in the Age of Artificial Intelligence”</a>, published in 2017 by a MIT physicist, Max Tegmark. The author included an image illustrating what computers (including AI) were able to do, where elevation represents difficulty for computers. As expected, arithmetic is located at the bottom of the elevation — i.e it’s a relatively easy task for computers to do. On the other hand, areas like science or theorem proving are more difficult. I find it most interesting that things like art, book writing, and programming were considered difficult for computers to do in 2017, but are now increasingly easier tasks to accomplish with advancements in large language models (LLMs).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/574/0*Ib6ITYzGoyxeKTBp.png" /><figcaption>Hans Moravec’s “landscape of human competence”</figcaption></figure><p>Below are some examples of how we can use AI to create impressive art.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VSo2FCGnEm0Bc8v2I7V8gg.png" /><figcaption>Random artwork generated by the Imagen 4 model.</figcaption></figure><p>In August 2022, an AI-generated image called “Théâtre D’opéra Spatial” secured first place in the digital category at the Colorado State Fair. This caused outrage <a href="https://www.nytimes.com/2022/09/02/technology/ai-artificial-intelligence-artists.html">amongst artists worldwide</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-lFyTA-gHjYn-QQE.jpeg" /><figcaption>“Théâtre D’opéra Spatial” generated by Midjourney AI</figcaption></figure><p>In April 2024, OpenAI created a short film titled “<a href="https://www.techradar.com/computing/artificial-intelligence/turns-out-the-viral-air-head-sora-video-wasnt-purely-the-work-of-ai-we-were-led-to-believe">air head</a>”. The video went viral — mainly because of the fact that the scenes in the video were generated by a text-to-video model SORA, which is able to produce highly realistic video content. The raw video outputs were then cleaned and processed to produce the final video below.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F9oryIMNVtto%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D9oryIMNVtto&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F9oryIMNVtto%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/77d2f2d1b9c26ecacfd284ed5632cde0/href">https://medium.com/media/77d2f2d1b9c26ecacfd284ed5632cde0/href</a></iframe><p>A major breakthrough in text-to-video happened in May 2025 when Google released the Veo3 text-to-video model. This model is able to generate short videos that include corresponding audio purely from a prompt. An example of a fictional video from Veo3 is shown below.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FOiuJfZ7LX1c%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DOiuJfZ7LX1c&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FOiuJfZ7LX1c%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/e73d84b7ef218f160d9645cf2b44d2bb/href">https://medium.com/media/e73d84b7ef218f160d9645cf2b44d2bb/href</a></iframe><h3>AI in book writing</h3><p>A number of books have now been written by authors actively using GenAI as part of their writing process. Authors actively use GenAI to structure their content, write drafts, rewrite drafts to adapt them to a particular style, explain certain concepts, review early versions, and more!</p><p>In addition, complete books have now been written by LLMs. Examples include:</p><ul><li><a href="https://amzn.to/4l81KEJ">1 the Road</a></li><li><a href="https://amzn.to/44HHy6L">The Aum Golly Series — Poems on Humanity by Artificial Intelligence</a></li><li><a href="https://amzn.to/4nuC1b5">The Serious: A Proven Divorce</a></li><li><a href="https://amzn.to/4l0sPts">Bob The Robot: Exploring the Universe — A Cozy Bedtime Story Produced by Artificial Intelligence</a></li><li><a href="https://amzn.to/40zh3Oq">The Inner Life of an AI: A Memoir by ChatGPT</a></li></ul><h3>AI in programming</h3><p>One of the great benefits of GenAI technologies is the ability to assist with the programming process. Programmers are now able to become more efficient by collaborating with LLM-based tools like <a href="https://github.com/features/copilot">Github Copilot</a>, <a href="https://www.anthropic.com/claude-code">Claude Code</a>, or fully outsource building applications to AI by following the <a href="https://en.wikipedia.org/wiki/Vibe_coding">“vibe coding”</a> approach and using tools like <a href="https://www.cursor.com/">Cursor</a>, <a href="https://devin.ai/">Devin</a>, <a href="https://v0.dev/">v0</a>, <a href="https://windsurf.com/">Windsurf</a>, or <a href="https://jules.google/">Jules </a>from Google.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FfjHtjT7GO1c%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DfjHtjT7GO1c&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FfjHtjT7GO1c%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/6c8f59e40b2565073c093780c7309dd5/href">https://medium.com/media/6c8f59e40b2565073c093780c7309dd5/href</a></iframe><h3>Conclusion</h3><p>The year is 2025, and AI has undeniably defied expectations. Just a few years ago, we considered creative tasks like art, writing, and even programming to be beyond machines. Yet, as we’ve explored, AI is now generating award-winning art, crafting realistic videos, writing entire books, and acting as a fully-fledged software engineer. This rapid progress has shattered previous assumptions about what AI can do. The “landscape of human competence” has been dramatically reshaped, proving that the boundaries of AI’s capabilities are far more fluid and surprising than we ever predicted. The future of AI is unfolding beyond our expectations, and it’s just getting started.</p><p>If you are considering starting a new Generative AI initiative but find it hard to make sense of the ever-changing AI landscape, we at Credera have vast experience in implementing AI solutions for our clients, and we’ll be more than happy to help you. You can find more information on our website: <a href="https://www.credera.co.uk/en-gb/services/artificial-intelligence">Credera AI</a></p><h3>Got a question?</h3><p>Please <a href="https://www.credera.com/en-us/services/artificial-intelligence">get in touch</a> to speak to a member of our team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7aa1c0d29aca" width="1" height="1" alt=""><hr><p><a href="https://medium.com/credera-engineering/beyond-our-expectations-how-ai-defied-predictions-7aa1c0d29aca">Beyond expectations: How AI has defied predictions</a> was originally published in <a href="https://medium.com/credera-engineering">Credera Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Here’s Everything You Need to Know About the Bank of England Base Rate Change and How It Affects…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@konradbachusz/heres-everything-you-need-to-know-about-the-bank-of-england-base-rate-change-and-how-it-affects-b94d9d307027?source=rss-6ed77e3d609------2"><img src="https://cdn-images-1.medium.com/max/850/0*qSHoRKusrI3PKi4I.jpg" width="850"></a></p><p class="medium-feed-snippet">Introduction</p><p class="medium-feed-link"><a href="https://medium.com/@konradbachusz/heres-everything-you-need-to-know-about-the-bank-of-england-base-rate-change-and-how-it-affects-b94d9d307027?source=rss-6ed77e3d609------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@konradbachusz/heres-everything-you-need-to-know-about-the-bank-of-england-base-rate-change-and-how-it-affects-b94d9d307027?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/b94d9d307027</guid>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[united-kingdom]]></category>
            <category><![CDATA[personal-finance]]></category>
            <category><![CDATA[news]]></category>
            <category><![CDATA[finance]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Fri, 04 Jul 2025 09:52:10 GMT</pubDate>
            <atom:updated>2025-07-04T13:56:43.619Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Introduction to GenAI-assisted code reviews]]></title>
            <link>https://medium.com/credera-engineering/introduction-to-genai-assisted-code-reviews-7237f6fab79d?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/7237f6fab79d</guid>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Fri, 09 May 2025 09:12:53 GMT</pubDate>
            <atom:updated>2025-05-09T10:39:58.989Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*xrDBMdaRYOdzkGR9.jpg" /><figcaption>2024 Crowdstrike Outage: Thorough code reviews lead to fewer bugs</figcaption></figure><h3><strong>Introduction</strong></h3><p>During the July 2024 <a href="https://www.techtarget.com/whatis/feature/Explaining-the-largest-IT-outage-in-history-and-whats-next">CrowdStrike outage</a>, over 8.5 million systems crashed and were unable to properly restart. The outage affected organisations like airlines, airports, banks, hotels, hospitals, manufacturing, stock markets, broadcasting, gas stations, retail stores, and governmental services. Imagine being at an airport, hopelessly staring at a series of “blue screens of death” while finding out that your flight was cancelled because of some unidentified software bug…</p><p>It is estimated that it led to worldwide damage of up to <strong>US$10 billion!</strong> The exact details of the outage can be found <a href="https://en.wikipedia.org/wiki/2024_CrowdStrike-related_IT_outages">here</a>. However, it is important to emphasise the importance of verifying, reviewing, and testing code quality to prevent such events.</p><p>This article will highlight the significance of code reviews and show you how to improve your efficiency with the help of GenAI.</p><h3>Code reviews</h3><p>Code reviews are an essential aspect of the software development lifecycle that are a cornerstone of building high-quality, maintainable, and reliable software. They are a systematic examination of source code intended to identify potential issues, improve code quality, ensure adherence to coding standards, and facilitate knowledge sharing within the team.</p><p>There are several benefits associated with code reviews</p><ul><li><strong>Improved code quality:</strong> Multiple sets of eyes on the code significantly increase the chances of catching bugs, logical errors, and edge cases that the original author might have missed. This leads to more robust and stable software.</li><li><strong>Early bug detection:</strong> Identifying issues early in the development cycle is far more cost-effective than fixing them in later stages or in production. Code reviews act as a preventative measure. This is imperative in the context of failures such as the <a href="https://www.techtarget.com/whatis/feature/Explaining-the-largest-IT-outage-in-history-and-whats-next">CrowdStrike outage</a> in July 2024.</li><li><strong>Knowledge sharing and learning:</strong> Reviews provide a platform for team members to learn from each other. Less experienced developers can pick up best practices and new techniques from senior engineers, while even experienced developers can gain fresh perspectives and discover alternative solutions.</li><li><strong>Enforcing coding standards and consistency:</strong> Code reviews ensure that the codebase adheres to established coding standards and style guides. This consistency improves readability and maintainability, making it easier for the entire team to work on the code.</li><li><strong>Reduced technical debt:</strong> By identifying and addressing suboptimal code, potential performance bottlenecks, and architectural flaws early on, code reviews help prevent the accumulation of technical debt, which can hinder future development efforts.</li><li><strong>Enhanced team collaboration and communication:</strong> The review process fosters discussion and collaboration among team members. It encourages open communication about code design and implementation choices.</li><li><strong>Improved code maintainability:</strong> Code that is well-reviewed is generally more readable, understandable, and consistently structured, making it easier to maintain and modify in the future.</li><li><strong>Onboarding and knowledge transfer:</strong> Code reviews help new team members quickly become familiar with the codebase and the team&#39;s coding practices. They also prevent knowledge silos by ensuring that multiple developers understand different parts of the system.</li><li><strong>Security vulnerability detection:</strong> Reviews can help identify potential security vulnerabilities that may have been overlooked during the initial development.</li></ul><p>It is a highly manual process and requires a lot of expertise from the reviewers. In the next section, I’ll show you how you can use GenAI to review code to make improvements and identify vulnerabilities.</p><h3>GenAI-assisted code reviews</h3><p>To demonstrate how GenAI can assist with code reviews, I’ve written a piece of poor-quality code. It includes several issues, such as excessive unnecessary imports, hardcoded credentials, a cluttered single function, SQL injection vulnerabilities, insecure data handling, and more.</p><p>See the example below:</p><pre>import psycopg2, sys, os, hashlib, base64, pickle, json, csv, datetime, math, random, socket, urllib.request<br>from collections import defaultdict, OrderedDict<br>from functools import reduce<br>from itertools import chain<br><br>def do_everything_terribly():<br>    creds = {<br>        &#39;host&#39;: &#39;prod-db.example.com&#39;,<br>        &#39;dbname&#39;: &#39;users&#39;,<br>        &#39;user&#39;: &#39;superadmin&#39;,<br>        &#39;password&#39;: &#39;Admin123!&#39;,<br>        &#39;port&#39;: &#39;5432&#39;<br>    }<br>    <br>    conn = psycopg2.connect(**creds)<br>    cur = conn.cursor()<br>    <br>    user_id = input(&quot;Enter user ID to search: &quot;)<br>    cur.execute(f&quot;SELECT * FROM users WHERE id = {user_id}&quot;)<br>    <br>    users = cur.fetchall()<br>    user_emails = [u[2] for u in users]<br>    <br>    with open(&#39;creds.json&#39;, &#39;w&#39;) as f:<br>        json.dump(creds, f)<br>    with open(&#39;creds.pickle&#39;, &#39;wb&#39;) as f:<br>        pickle.dump(creds, f)<br>    <br>    print(&quot;USER CREDENTIALS:&quot;)<br>    for u in users:<br>        print(f&quot;User: {u[1]}, Pass: {u[3]}&quot;)<br>    <br>    fake_hashes = [hashlib.md5(pwd.encode()).hexdigest() for _, _, _, pwd, _ in users]<br>    <br>    with open(&#39;report.csv&#39;, &#39;w&#39;) as f:<br>        writer = csv.writer(f)<br>        writer.writerow([&#39;Username&#39;, &#39;Password&#39;, &#39;FakeHash&#39;])<br>        for u, h in zip(users, fake_hashes):<br>            writer.writerow([u[1], u[3], h])<br>    <br>    return {<br>        &#39;users&#39;: users,<br>        &#39;hashes&#39;: fake_hashes,<br>        &#39;connection&#39;: conn<br>    }<br><br>if __name__ == &quot;__main__&quot;:<br>    result = do_everything_terribly()<br>    globals().update(result)</pre><p>I can pass it to a GenAI tool of choice while including a prompt such as the one shown below:</p><pre>You are a senior engineer. Please review this code and offer suggestions. <br>Think step by step:<br>&lt;your_code&gt;</pre><p>This would generally work with different GenAI tools like ChatGPT, Gemini, DeepSeek, Microsoft Copilot, amongst others.</p><p>The Large Language Model (LLM) will review your code and give you some feedback on how you can improve it. This will include identified issues and improvement suggestions. The example from DeepSeek is shown below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/647/1*-Hpd4VL_kpJJ2-1U3Y4bkQ.png" /><figcaption>GenAI-assisted code review output example</figcaption></figure><p>You see output as a series of potential issues with your code, followed by suggestions on how to improve it. This is very helpful in terms of getting a quick “scan” of your code and spotting any issues that you may have missed. Addressing many of those would lead to improvements in the quality and security of your code.</p><h3>Tread with caution</h3><p>It is important to state that you should use this approach with caution when dealing with sensitive codebases and never paste sensitive data or secrets (e.g passwords and API tokens). This approach is ideal if you can ensure that your GenAI tool doesn’t share your data with third parties or if you’re running your LLM locally. Consider using environment variables instead of hardcoding secrets.</p><p>GenAI-assisted code reviews can lead to significant efficiency gains. However, as great as it sounds, we shouldn’t outsource code reviews to AI completely. As an experienced engineer, you should always use your judgment about which AI-generated code review feedback is valid.</p><h3>Conclusion</h3><p>In this article, we explored how GenAI can assist with code reviews, helping to identify vulnerabilities that are easy to overlook while offering suggestions for improving the codebase. I hope this guide helps you and your team work more efficiently and continuously enhance the quality of your code!</p><h3>Got questions?</h3><p>If you are considering starting a new Generative AI initiative but find it hard to make sense of the ever-changing AI landscape, we at Credera have vast experience in helping our clients generate enterprise value from their data.</p><p>Please <a href="https://hubs.la/Q012VcRY0">get in touch</a> to speak to a member of our team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7237f6fab79d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/credera-engineering/introduction-to-genai-assisted-code-reviews-7237f6fab79d">Introduction to GenAI-assisted code reviews</a> was originally published in <a href="https://medium.com/credera-engineering">Credera Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Code smarter, not harder: The developer’s introduction to AI-powered coding.]]></title>
            <link>https://medium.com/credera-engineering/code-smarter-not-harder-the-developers-introduction-to-ai-powered-coding-909341fd3297?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/909341fd3297</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[productivity]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Thu, 24 Apr 2025 10:11:34 GMT</pubDate>
            <atom:updated>2025-04-24T10:11:34.768Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZgV2dioxb75w-QESuPFv6w.png" /><figcaption>“Vibe Coding” in action…</figcaption></figure><h4>Introduction</h4><blockquote>“AI won’t take your job. It’s somebody using AI that will take your job.”</blockquote><blockquote>Richard Baldwin</blockquote><p>The above statement illustrates the dramatic impact of AI technologies on society. The increasing popularity of Generative AI (GenAI) allows developers to become significantly more productive. This article will teach you how you can use GenAI to become 10x more productive in writing code. Let’s deep dive into some specific use cases.</p><h4>Writing code</h4><p>Because many large language models were trained on a corpus of data that contains a large amount of code, they are good at generating new lines of code. This is particularly useful when it comes to auto-completion, generating code drafts, or writing simpler programmes.</p><p>We can do this using tools with a chatbot interface, e.g., ChatGPT, Gemini, DeepSeek, etc. In the example below, we ask to write some simple Rust code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*tBwAAXVJFzoZPrvP1YiM7w.png" /><figcaption>Example Rust code generated by AI</figcaption></figure><p>We can test that the code executes correctly and gives us what we expect.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MBWzmavHoxu-DotsbWZHnw.png" /><figcaption>Executing Rust code to calculate the value of Pi</figcaption></figure><p>We can also do it directly within your IDE with tools like <a href="https://marketplace.visualstudio.com/items?itemName=Google.geminicodeassist">Gemini Code Assist</a>, <a href="https://www.qodo.ai/products/ide-plugin/">Qodo Gen</a>, or <a href="https://github.com/features/copilot">Github Copilot</a>. This allows us to autocomplete our code or ask for what we want directly in the codebase. The advantage of this is that often the LLM would have the awareness of other files within your codebase. The Gemini example below shows how we can do this.</p><p><strong>Initial prompt</strong></p><pre>#Write a Python Function that converts a JSON object to a pandas dataframe. <br>Explain every step.</pre><p><strong>Output in VSCode:</strong></p><pre>#Write a Python Function that converts a JSON object to a pandas dataframe. Explain every step.<br>import pandas as pd<br>import json<br><br>def json_to_dataframe(json_data):<br>    &quot;&quot;&quot;<br>    Converts a JSON object to a pandas DataFrame.<br><br>    Args:<br>        json_data: A JSON object (string or dictionary).<br><br>    Returns:<br>        A pandas DataFrame representing the JSON data, or None if an error occurs.<br>    &quot;&quot;&quot;<br>    try:<br>        # Step 1: Check if the input is a string or a dictionary.<br>        if isinstance(json_data, str):<br>            # Step 2: If it&#39;s a string, parse it as JSON.<br>            data = json.loads(json_data)<br>        elif isinstance(json_data, dict):<br>            # Step 3: If it&#39;s a dictionary, use it directly.<br>            data = json_data<br>        else:<br>            raise TypeError(&quot;Input must be a JSON string or a dictionary.&quot;)<br><br>        # Step 4: Check if the parsed data is a list of dictionaries or a single dictionary.<br>        if isinstance(data, list):<br>            # Step 5: If it&#39;s a list, assume each element is a row.<br>            df = pd.DataFrame(data)<br>        elif isinstance(data, dict):<br>            # Step 6: If it&#39;s a dictionary, treat it as a single row.<br>            df = pd.DataFrame([data])<br>        else:<br>            raise ValueError(&quot;JSON data must be a dictionary or a list of dictionaries.&quot;)<br><br>        # Step 7: Return the DataFrame.<br>        return df<br><br>    except json.JSONDecodeError as e:<br>        print(f&quot;Error decoding JSON: {e}&quot;)<br>        return None<br>    except TypeError as e:<br>        print(f&quot;Type Error: {e}&quot;)<br>        return None<br>    except ValueError as e:<br>        print(f&quot;Value Error: {e}&quot;)<br>        return None</pre><p>The second prompt is provided to call the generated function:</p><pre>#Call json_to_dataframe with dummy JSON</pre><p>The output code calls the Python function generated in the previous step.</p><pre>#Call json_to_dataframe with dummy JSON<br>dummy_json_string = &#39;[{&quot;name&quot;: &quot;Alice&quot;, &quot;age&quot;: 30, &quot;city&quot;: &quot;New York&quot;}, {&quot;name&quot;: &quot;Bob&quot;, &quot;age&quot;: 25, &quot;city&quot;: &quot;Los Angeles&quot;}]&#39;<br>dummy_json_dict = {&quot;name&quot;: &quot;Charlie&quot;, &quot;age&quot;: 35, &quot;city&quot;: &quot;Chicago&quot;}<br><br>df_from_string = json_to_dataframe(dummy_json_string)<br>if df_from_string is not None:<br>    print(&quot;DataFrame from string:&quot;)<br>    print(df_from_string)<br><br>df_from_dict = json_to_dataframe(dummy_json_dict)<br>if df_from_dict is not None:<br>    print(&quot;\nDataFrame from dictionary:&quot;)<br>    print(df_from_dict)</pre><p>You can perform this process in the IDE of your choice to speed up the development process. In this example, I used VSCode and the <a href="https://marketplace.visualstudio.com/items?itemName=Google.geminicodeassist">Gemini Code Assist</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5E_EpjOuij1Rc6e0Fe_2UA.png" /><figcaption>Assisted coding in VSCode</figcaption></figure><p>More recent tools, such as the ones below, include the agentic mode, which would break your request into a set of tasks that would be actioned by an AI agent.</p><ul><li><a href="https://www.cursor.com/">Cursor</a></li><li><a href="https://v0.dev/">v0</a></li><li><a href="https://bolt.new/">Bolt</a></li><li><a href="https://docs.windsurf.com/windsurf/getting-started">Windsurf</a></li><li><a href="https://firebase.studio/">Firebase Studio</a></li></ul><p>The choice depends on your preference and the limitations of some of those tools, as well as how much control you want to have as opposed to just fully embracing “<a href="https://en.wikipedia.org/wiki/Vibe_coding">vibe coding</a>”. The example below shows a more complex task where I asked it to create a “Tinder for Cats”. I used the tool called <a href="https://v0.dev/">v0</a> to do it for me. Here’s some AI-generated TypeScript code:</p><pre>import Link from &quot;next/link&quot;<br>import Image from &quot;next/image&quot;<br>import { Button } from &quot;@/components/ui/button&quot;<br>import { Card, CardContent, CardFooter } from &quot;@/components/ui/card&quot;<br>import { MessageCircle } from &quot;lucide-react&quot;<br><br>// Sample matches data with real images<br>const matches = [<br>  {<br>    id: 2,<br>    name: &quot;Luna&quot;,<br>    age: 2,<br>    breed: &quot;Siamese&quot;,<br>    lastActive: &quot;5 minutes ago&quot;,<br>    image: &quot;/images/cat2.jpg&quot;,<br>  },<br>  {<br>    id: 4,<br>    name: &quot;Bella&quot;,<br>    age: 1,<br>    breed: &quot;Ragdoll&quot;,<br>    lastActive: &quot;2 hours ago&quot;,<br>    image: &quot;/images/cat4.jpg&quot;,<br>  },<br>]<br><br>export default function MatchesPage() {<br>  return (<br>    &lt;div className=&quot;flex flex-col min-h-screen&quot;&gt;<br>      &lt;header className=&quot;border-b&quot;&gt;<br>        &lt;div className=&quot;container flex items-center justify-between py-4&quot;&gt;<br>          &lt;Link href=&quot;/&quot; className=&quot;flex items-center gap-2&quot;&gt;<br>            &lt;span className=&quot;text-2xl font-bold bg-gradient-to-r from-pink-500 to-orange-500 bg-clip-text text-transparent&quot;&gt;<br>              Whisker Match<br>            &lt;/span&gt;<br>          &lt;/Link&gt;<br>          &lt;nav className=&quot;flex items-center gap-4&quot;&gt;<br>            &lt;Link href=&quot;/matches&quot;&gt;<br>              &lt;Button variant=&quot;ghost&quot; size=&quot;sm&quot; className=&quot;font-bold&quot;&gt;<br>                Matches<br>              &lt;/Button&gt;<br>            &lt;/Link&gt;<br>            &lt;Link href=&quot;/profile&quot;&gt;<br>              &lt;Button variant=&quot;ghost&quot; size=&quot;sm&quot;&gt;<br>                Profile<br>              &lt;/Button&gt;<br>            &lt;/Link&gt;<br>          &lt;/nav&gt;<br>        &lt;/div&gt;<br>      &lt;/header&gt;<br>      &lt;main className=&quot;flex-1 container py-6&quot;&gt;<br>        &lt;h1 className=&quot;text-3xl font-bold mb-8&quot;&gt;Your Matches&lt;/h1&gt;<br><br>        {matches.length === 0 ? (<br>          &lt;div className=&quot;text-center py-12&quot;&gt;<br>            &lt;p className=&quot;text-muted-foreground mb-4&quot;&gt;You don&#39;t have any matches yet.&lt;/p&gt;<br>            &lt;Link href=&quot;/&quot;&gt;<br>              &lt;Button&gt;Start Swiping&lt;/Button&gt;<br>            &lt;/Link&gt;<br>          &lt;/div&gt;<br>        ) : (<br>          &lt;div className=&quot;grid grid-cols-1 md:grid-cols-2 lg:grid-cols-3 gap-6&quot;&gt;<br>            {matches.map((match) =&gt; (<br>              &lt;Card key={match.id}&gt;<br>                &lt;div className=&quot;relative pt-[100%]&quot;&gt;<br>                  &lt;Image<br>                    src={match.image || &quot;/placeholder.svg&quot;}<br>                    alt={match.name}<br>                    fill<br>                    sizes=&quot;(max-width: 768px) 100vw, 300px&quot;<br>                    className=&quot;object-cover rounded-t-lg&quot;<br>                    onError={(e) =&gt; {<br>                      const target = e.target as HTMLImageElement<br>                      target.src = &quot;/placeholder.svg?height=200&amp;width=200&quot;<br>                    }}<br>                  /&gt;<br>                &lt;/div&gt;<br>                &lt;CardContent className=&quot;p-4&quot;&gt;<br>                  &lt;h2 className=&quot;text-xl font-bold&quot;&gt;<br>                    {match.name}, {match.age}<br>                  &lt;/h2&gt;<br>                  &lt;p className=&quot;text-sm text-muted-foreground&quot;&gt;{match.breed}&lt;/p&gt;<br>                  &lt;p className=&quot;text-xs text-muted-foreground mt-1&quot;&gt;Active {match.lastActive}&lt;/p&gt;<br>                &lt;/CardContent&gt;<br>                &lt;CardFooter className=&quot;p-4 pt-0&quot;&gt;<br>                  &lt;Button className=&quot;w-full&quot; size=&quot;sm&quot;&gt;<br>                    &lt;MessageCircle className=&quot;h-4 w-4 mr-2&quot; /&gt;<br>                    Message<br>                  &lt;/Button&gt;<br>                &lt;/CardFooter&gt;<br>              &lt;/Card&gt;<br>            ))}<br>          &lt;/div&gt;<br>        )}<br>      &lt;/main&gt;<br>      &lt;footer className=&quot;border-t py-4&quot;&gt;<br>        &lt;div className=&quot;container text-center text-sm text-muted-foreground&quot;&gt;<br>          © {new Date().getFullYear()} Whisker Match. All rights reserved.<br>        &lt;/div&gt;<br>      &lt;/footer&gt;<br>    &lt;/div&gt;<br>  )<br>}</pre><p>Corresponding output:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z6P9reLgNdYaUAwHTJTi_Q.png" /><figcaption>Tinder for cats created by GenAI</figcaption></figure><h4>Conclusion</h4><p>This article shows you how you can use GenAI to convert your ideas to working code. While GenAI can greatly assist in writing code, it is prone to mistakes and hallucinations. To understand more about the challenges of Generative AI technology, check out my other article:</p><p><a href="https://medium.com/credera-engineering/use-cases-and-challenges-of-generative-ai-for-text-data-fcaf4f13322b">Use cases and challenges of Generative AI for text data</a></p><p>At the end of the day, it is <em>your</em> responsibility to use your experience and judgment to decide which suggestions from the LLM you are going to accept. <strong>Blaming AI for code errors is like blaming the compiler for a production outage.</strong></p><p>In the upcoming months, we are likely to see a lot of new tools that help individuals build exciting stuff using GenAI technology!</p><p>I hope you learned something useful. Stay tuned for more technology productivity and AI &amp; data content in my upcoming articles!</p><h3>Got a question?</h3><p>Please <a href="https://hubs.la/Q012VcRY0">get in touch</a> to speak to a member of our team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=909341fd3297" width="1" height="1" alt=""><hr><p><a href="https://medium.com/credera-engineering/code-smarter-not-harder-the-developers-introduction-to-ai-powered-coding-909341fd3297">Code smarter, not harder: The developer’s introduction to AI-powered coding.</a> was originally published in <a href="https://medium.com/credera-engineering">Credera Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Top ten tips for passing the AWS Certified Cloud Practitioner Certification exam]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/credera-engineering/top-ten-tips-for-passing-the-aws-certified-cloud-practitioner-certification-exam-671f73d7defc?source=rss-6ed77e3d609------2"><img src="https://cdn-images-1.medium.com/max/1000/0*wlMFfRRvAyTlLTmD" width="1000"></a></p><p class="medium-feed-snippet">In this blog, Engineer Konrad Baschusz sets out his ten top tips for acing the AWS Certified Cloud Practitioner certification exam.</p><p class="medium-feed-link"><a href="https://medium.com/credera-engineering/top-ten-tips-for-passing-the-aws-certified-cloud-practitioner-certification-exam-671f73d7defc?source=rss-6ed77e3d609------2">Continue reading on Credera Engineering »</a></p></div>]]></description>
            <link>https://medium.com/credera-engineering/top-ten-tips-for-passing-the-aws-certified-cloud-practitioner-certification-exam-671f73d7defc?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/671f73d7defc</guid>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Mon, 29 Jan 2024 09:29:06 GMT</pubDate>
            <atom:updated>2025-05-04T11:43:32.716Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Seven reasons why technology leaders are paying attention to OpenAI’s DevDay]]></title>
            <link>https://medium.com/credera-engineering/seven-reasons-why-technology-leaders-are-paying-attention-to-openais-devday-fa908e3e8e6f?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/fa908e3e8e6f</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[generative-ai-use-cases]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Tue, 14 Nov 2023 15:30:44 GMT</pubDate>
            <atom:updated>2023-11-14T15:30:44.074Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jVBBycBK5znLN2rMGqet7Q.jpeg" /><figcaption>Photo by Hatice Baran: <a href="https://www.pexels.com/photo/open-laptop-on-desk-16037283/">https://www.pexels.com/photo/open-laptop-on-desk-16037283/</a></figcaption></figure><p>On 6th November 6 2023, the city of San Francisco hosted the highly anticipated OpenAI DevDay - an immersive event that brought together innovators, developers, and enthusiasts at the forefront of Generative AI. This was an iPhone moment for AI as a field, and given the rapid development of the area, it’s important to stay up-to-date.</p><p>This blog highlights the major announcements from the event.</p><h3><strong>1. GPT-4 Turbo model</strong></h3><p>A new GPT model was made available on the OpenAI platform. It supports the following:</p><ul><li><strong>Context Length:</strong> GPT-4 Turbo supports a context window of up to 128,000 tokens, which is equivalent to around 300 pages of text, enabling it to process much larger amounts of information in a single prompt.</li><li><strong>Knowledge Cutoff:</strong> The model has knowledge of events up to April 2023, ensuring that its responses are up-to-date with recent developments.</li><li><strong>Performance:</strong> It is more capable than the standard GPT-4 model, with optimised performance for better task handling.</li><li><strong>Cost:</strong> GPT-4 Turbo is significantly cheaper, costing three times less for input tokens and two times less for output tokens compared to the standard GPT-4 model.</li><li><strong>JSON Mode: </strong>New JSON mode ensures that the model’s responses are in valid JSON. The response_format API parameter allows users to constrain the model’s output, ensuring the generation of syntactically correct JSON objects. This feature is particularly useful for developers using the Chat Completions API to generate JSON outside of function calling.</li><li><strong>Reproducibility:</strong> A new seed parameter allows for reproducible outputs, providing consistent completions that are particularly useful for debugging and unit testing. <a href="https://platform.openai.com/docs/guides/text-generation/reproducible-outputs">Learn more about reproducible outputs here.</a> In the coming weeks, OpenAI will introduce a functionality that provides log probabilities for the most probable output tokens generated by both GPT-4 Turbo and GPT-3.5 Turbo.</li><li><strong>Multimodal Capabilities:</strong> GPT-4 Turbo can accept images as inputs, expanding its use to tasks like generating captions, analysing images, reading documents with figures, creating images, and transforming text to speech.</li></ul><p>Developers with a paid subscription now have the opportunity to experiment with GPT-4 Turbo by passing gpt-4–1106-preview<strong> </strong>in the API. OpenAI is on track to launch the stable, production-ready model in the upcoming weeks.</p><h3>2. GPTs</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7jroXo21cEVp3ni-" /><figcaption>OpenAI GPTs</figcaption></figure><p>OpenAI has introduced GPTs, allowing users to create custom versions of ChatGPT tailored for specific purposes, enhancing its utility in daily life, tasks, work, or home. GPTs enable easy customisation without coding, allowing users to build and share their creations, such as learning board game rules, teaching maths, or designing stickers. Users can try GPTs available for ChatGPT Plus and Enterprise users, including Canva and Zapier AI Actions. OpenAI plans to launch the GPT Store later this month, showcasing creations by verified builders. Privacy and safety measures are emphasised, with user control over data and systems in place to review GPTs against usage policies. Developers can connect GPTs to the real world through custom actions, and enterprises can deploy internal-only GPTs for specific use cases. OpenAI emphasises community involvement in shaping AI behaviour and announces updates to ChatGPT Plus for fresher information and a simplified user experience.</p><p>Creating one is as easy as starting a chat, where users can provide instructions and extra knowledge, and choose what it can do, like making images, searching the web, or analysing data. Try it out at <a href="https://chat.openai.com/create">https://chat.openai.com/create</a></p><h3><strong>3. Assistants API:</strong></h3><p>This marks a significant step in facilitating developers to create agent-like experiences within their applications. This API empowers purpose-built AI assistants with specific instructions, additional knowledge, and the ability to call models and tools for task execution. The API beta is available for exploration on the <a href="https://platform.openai.com/playground">Assistants playground</a>, without the need for coding.</p><h3><strong>4. Code Interpreter:</strong></h3><p>Code Interpreter writes and runs Python code in a sandboxed execution environment, generating graphs and charts, and processing files with diverse data and formatting. It allows your assistants to run code iteratively to solve challenging code, maths problems, and more.</p><h3><strong>5. Built-in Retrieval</strong></h3><p>Built-in retrieval augments the assistant with knowledge from outside our models, such as proprietary domain data, product information, or documents provided by your users. This means you don’t need to compute and store embeddings for your documents or implement chunking and search algorithms. The Assistants API optimises what retrieval technique to use based on our experience building knowledge retrieval in ChatGPT.</p><h3>6. Function calling updates</h3><p>The function calling feature enables the description of your app or external APIs’ functions to models, allowing the model to intelligently generate a JSON object containing arguments for calling those functions. In the latest updates, users can now request multiple actions in a single message, streamlining the process. For example, a user can send a message like “open the car window and turn off the A/C” in one go, which previously would have required multiple interactions with the model.</p><h3>7. Lower usage prices and increased rate limits</h3><p>A reduction in prices (per 1,000 tokens) has been introduced to make the platform more accessible to developers. In addition, the tokens per minute limit has been doubled to help with scaling your applications. The current pricing is shown below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/923/1*XCRb2L7M39v4AXSN0jpeag.png" /><figcaption>OpenAI November 2023 pricing</figcaption></figure><h3>Summary</h3><ul><li><strong>GPT-4 Turbo</strong>: Enhanced version with a larger context window, up-to-date knowledge, optimised performance, lower costs, JSON mode, reproducibility, multimodal capabilities, and a preview available for developers.</li><li><strong>GPTs Introduction</strong>: Users can create custom ChatGPT versions for specific tasks without coding, with a GPT Store launching soon.</li><li><strong>Assistants API</strong>: New API for building AI assistants with specific capabilities in applications, available for beta testing.</li><li><strong>Code Interpreter</strong>: A tool for writing and running Python code with support for graphs, charts, and diverse data processing.</li><li><strong>Built-in Retrieval</strong>: Enhances the assistant with external knowledge without the need for complex search algorithms.</li><li><strong>Function Calling Updates</strong>: Allows for multiple actions in a single message and improved function calling accuracy.</li><li><strong>Lower Usage Prices and Increased Rate Limits</strong>: Reduced prices and higher token rate limits to support developers.</li></ul><p>If you are exploring Generative AI within your organisation and would like to hear about Credera’s wealth of experience in implementing Generative AI to increase the productivity of clients, please don’t hesitate to get in touch. You can find more information on our website: <a href="https://www.credera.co.uk/ai-for-productivity">Credera AI</a></p><h3>Got a question?</h3><p>Please <a href="https://hubs.la/Q012VcRY0">get in touch</a> to speak to a member of our team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fa908e3e8e6f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/credera-engineering/seven-reasons-why-technology-leaders-are-paying-attention-to-openais-devday-fa908e3e8e6f">Seven reasons why technology leaders are paying attention to OpenAI’s DevDay</a> was originally published in <a href="https://medium.com/credera-engineering">Credera Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Use cases and challenges of Generative AI for text data]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/credera-engineering/use-cases-and-challenges-of-generative-ai-for-text-data-fcaf4f13322b?source=rss-6ed77e3d609------2"><img src="https://cdn-images-1.medium.com/max/2600/1*9dbKVuNtm8KuKRFesxtyxA.jpeg" width="6144"></a></p><p class="medium-feed-snippet">The aim of this article is to give you a better understanding of Generative AI technology.</p><p class="medium-feed-link"><a href="https://medium.com/credera-engineering/use-cases-and-challenges-of-generative-ai-for-text-data-fcaf4f13322b?source=rss-6ed77e3d609------2">Continue reading on Credera Engineering »</a></p></div>]]></description>
            <link>https://medium.com/credera-engineering/use-cases-and-challenges-of-generative-ai-for-text-data-fcaf4f13322b?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/fcaf4f13322b</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Wed, 01 Nov 2023 16:28:06 GMT</pubDate>
            <atom:updated>2023-11-01T16:28:06.432Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[AWS NLP Conference 2023 takeaways]]></title>
            <link>https://medium.com/credera-engineering/aws-nlp-conference-2023-takeaways-693ab9dd9fd2?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/693ab9dd9fd2</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Tue, 10 Oct 2023 11:10:38 GMT</pubDate>
            <atom:updated>2023-10-10T11:10:38.337Z</atom:updated>
            <content:encoded><![CDATA[<h3>AWS NLP Conference 2023: Key Takeaways</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Saz8YlWzGPLBPNOSMUlbHg.jpeg" /><figcaption>Photo by Lukas: <a href="https://www.pexels.com/photo/blue-retractable-pen-574070/">https://www.pexels.com/photo/blue-retractable-pen-574070/</a></figcaption></figure><h3>Introduction</h3><p>I recently attended the AWS NLP Conference 2023 in AWS’s London offices. The event is tailored for data specialists and technology leaders, with global leaders in Large Language Models (LLMs) and Natural Language Processing (NLP) in attendance. In this blog, I’m sharing some of my key takeaways from the event.</p><p>The majority of the talks, demos, and workshops were focused on LLMs and Generative AI. Given the popularity of LLMs and tools like ChatGPT, Github, Copilot, and Midjourney, Generative AI is currently considered to be in the Peak of Inflated Expectations stage of the <a href="https://www.gartner.com/en/articles/what-s-new-in-the-2023-gartner-hype-cycle-for-emerging-technologies">Gartner Hype Cycle for Emerging Technologies</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-9nyrbya2gXX20oAQ-K4yw.png" /><figcaption>Peak of Inflated Expectations</figcaption></figure><p>To find out more about common use cases and challenges of Generative AI in NLP please refer to my other blog post:</p><ul><li><a href="https://medium.com/@konradbachusz/fcaf4f13322b">Use Cases and Challenges of Generative AI for Text Data</a></li></ul><p>For now, let’s dive in to some of the key takeaways from the AWS NLP Conference 2023.</p><h3>Release of Falcon 180B language model</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-3a93_8CQmun5LxA.jpg" /></figure><p>On the 6th of September 2023, a new open-source language model was released by the <a href="https://www.tii.ae/">Technology Innovation Institute</a>. Falcon 180B is a powerful language model with 180 billion parameters, trained on 3.5 trillion tokens. It was trained using a staggering number of 4000 GPUs! It’s currently at the top of the <a href="https://www.maginative.com/article/falcon-180b-a-new-leader-in-open-llms-emerges/#:~:text=With%2068.74%20on%20the%20Hugging,Meta&#39;s%20LLaMA%202%20(67.35).">Hugging Face Leaderboard</a> for pre-trained Open Large Language Models and is available for both research and commercial use.</p><p>This model performs exceptionally well in a number of tasks such as reasoning, coding, proficiency, and knowledge tests - even beating competitors like Meta’s LLaMA 2. The model doesn’t use <a href="https://huggingface.co/blog/rlhf">Reinforcement learning from human feedback (RLHF)</a>.</p><p>Amongst closed-source models, it ranks just behind OpenAI’s GPT 4 and performs on par with Google’s PaLM 2 Large, which powers Bard, despite being half the size of the model.</p><p>Falcon-180B is accessible to developers through a royalty-free license, based on Apache 2.0. The licence includes restrictions on illegal or harmful use and requires those intending to provide hosted access to the model to seek additional consent from TII.</p><p>The model is free of charge to download, use, and integrate into applications and end-user products. You can access it <a href="https://huggingface.co/tiiuae/falcon-180B">here</a>.</p><p>Given the rapid development in the NLP space, Falcon 180B may be outperformed by a more powerful model this very moment.</p><h3>Model fine-tuning with Jumpstart</h3><p>AWS offers a solution known as Jumpstart that provides trained, open-source models for a wide range of problem types to help you get started with machine learning. You can incrementally train and tune these models before deployment. JumpStart also provides solution templates that set up infrastructure for common use cases, and executable example notebooks for machine learning with SageMaker.</p><p>A relatively new addition to the Jumpstart family are the Foundation Models.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/322/1*mv2oDVX1lPS-tDnHus1Krg.png" /></figure><p><a href="https://en.wikipedia.org/wiki/Foundation_models">Foundation models</a> are large-scale machine learning (ML) models that contain billions of parameters and are pre-trained on the vast datasets of text and image. You can perform a wide range of tasks such as article summarisation and text, image, or video generation. Because foundation models are pre-trained, they can help lower training and infrastructure costs and enable customisation for your use case. JumpStart supports popular proprietary foundation models, such as Llama-2 models, Falcon 180B, or Stable Diffusion. According to AWS, Jumpstart offers over 300 models.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*O6aU2X9nrQrLrbqM" /><figcaption>List of models available in SageMaker JumpStart</figcaption></figure><p>You can try out popular pre-trained foundation models without the need to deploy. To get started on the latest models that are in preview or to try out models in a playground, you need to request access. You’ll receive an email once it’s ready. If you have preview access, choose Foundation models under JumpStart in the <a href="https://console.aws.amazon.com/sagemaker/home?#/foundation-models">Amazon SageMaker console</a>. Playground lets you experiment with model capabilities via the AWS console.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*StkKhiPa4l5uQgD_" /><figcaption>AWS Console interface to send prompts to a model in the playground</figcaption></figure><p>More information on how to try the model in Playground mode, fine-tune it, or deploy it to production can be found in the links below:</p><ul><li><a href="https://community.aws/posts/using-proprietary-foundation-models-sagemaker-jumpstart-building-gen-ai-apps">Using Jumpstart Foundation Models</a></li><li><a href="https://github.com/aws/amazon-sagemaker-examples/blob/main/introduction_to_amazon_algorithms/jumpstart-foundation-models/domain-adaption-finetuning-gpt-j-6b.ipynb">Fine-tuning Jumpstart Models</a></li><li><a href="https://aws.amazon.com/blogs/machine-learning/question-answering-using-retrieval-augmented-generation-with-foundation-models-in-amazon-sagemaker-jumpstart/">Jumpstart with Retrieval Augmented Generation (RAG)</a></li></ul><h3>Introducing AWS Bedrock</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*vPCoMrBCoQLqgIPj" /></figure><p>Amazon Bedrock is a fully managed, serverless tool that makes foundation models available through an API, so you can choose from various foundation models to find the model that’s best suited for your use case. It was released in April this year and is yet to be made generally available. In the meantime, it is possible to get access to a beta release.</p><p>With the Amazon Bedrock serverless experience, you can quickly get started, easily experiment with FMs, privately customise FMs with your own data, and integrate and deploy them into your applications using AWS tools and capabilities.</p><p>Compared to the previously mentioned solution, SageMaker, JumpStart supports custom model training and utilises user-provided data. It is suitable for those who need to build models from scratch. Amazon Bedrock simplifies the process by leveraging pre-trained models, eliminating the need for custom training.</p><p>More information about Bedrock can be found here:</p><ul><li><a href="https://aws.amazon.com/bedrock/">Amazon Bedrock Official Page</a></li><li><a href="https://www.ml6.eu/blogpost/amazon-bedrock-building-a-solid-foundation-for-your-ai-strategy">Amazon Bedrock: Building a solid foundation for Your AI Strategy</a></li></ul><h3><strong>New AWS purpose-build accelerators for deep learning</strong></h3><p>Training, fine-tuning, or making inferences from LLMs can be time-consuming and expensive. Fortunately, AWS released a set of new compute accelerators to help with some of those challenges.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/0*KkQAHTDwJnd1xxsV.png" /><figcaption>AWS Trainium</figcaption></figure><ul><li><strong>AWS Trainium:</strong> AWS Trainium is the second-generation machine learning (ML) accelerator that AWS purpose-built for deep learning training of 100B+ parameter models. Each Amazon Elastic Compute Cloud (EC2) <a href="https://aws.amazon.com/ec2/instance-types/trn1/">Trn1 instance</a> deploys up to 16 AWS Trainium accelerators. Trainium-based EC2 Trn1 instances provide faster time to train whilst offering up to 50% cost-to-train savings over comparable Amazon EC2 instances. Trainium has been optimised for training NLP, computer vision, and recommender models.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*ggufh44lYoH6O8j8.png" /><figcaption>AWS Inferentia</figcaption></figure><ul><li><strong>AWS Inferentia:</strong> AWS Inferentia accelerators are designed by AWS to deliver high performance at the lowest cost for your deep learning (DL) inference applications. The first-generation AWS Inferentia accelerator powers <a href="https://aws.amazon.com/ec2/instance-types/inf1/">Amazon Elastic Compute Cloud (Amazon EC2) Inf1 instances</a>, which deliver up to 2.3x higher throughput and up to 70% lower cost per inference than comparable Amazon EC2 instances.</li></ul><figure><img alt="AWS Inferentia2" src="https://cdn-images-1.medium.com/max/374/0*4ROT9fTIQza7jnfq.png" /><figcaption>AWS Inferentia2</figcaption></figure><ul><li><strong>AWS Inferentia2: </strong>Amazon Elastic Compute Cloud (Amazon EC2) Inf2 instances are purpose-built for deep learning (DL) inference. They deliver high performance at the lowest cost in Amazon EC2 for generative artificial intelligence (AI) models, including large language models (LLMs) and vision transformers. Inf2 instances raise the performance of Inf1 by delivering 3x higher compute performance, 4x larger total accelerator memory, up to 4x higher throughput, and up to 10x lower latency.</li></ul><h3>Conclusion</h3><p>Despite the fact that the field is developing rapidly, we are able to see the emergence of useful tools, available via open source or large cloud vendors, that help organisations implement their AI solutions. I would highly recommend the AWS NLP Conference to anyone who is interested in AI, data, and NLP.</p><p>If you are considering starting a new Generative AI initiative but find it hard to make sense of the ever-changing AI landscape, we at Credera have vast experience in implementing AI solutions for our clients and we’ll be more than happy to help you. You can find more information on our website: <a href="https://www.credera.co.uk/en-gb/services/artificial-intelligence">Credera AI</a></p><h3>Got a question?</h3><p>Please <a href="https://hubs.la/Q012VcRY0">get in touch</a> to speak to a member of our team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=693ab9dd9fd2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/credera-engineering/aws-nlp-conference-2023-takeaways-693ab9dd9fd2">AWS NLP Conference 2023 takeaways</a> was originally published in <a href="https://medium.com/credera-engineering">Credera Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Benefits of Value Investing]]></title>
            <link>https://medium.com/@konradbachusz/benefits-of-value-investing-f91659dc5679?source=rss-6ed77e3d609------2</link>
            <guid isPermaLink="false">https://medium.com/p/f91659dc5679</guid>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[trading]]></category>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[konrad bachusz]]></dc:creator>
            <pubDate>Sun, 13 Nov 2022 10:52:56 GMT</pubDate>
            <atom:updated>2022-11-13T10:52:56.694Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7i6Tmc56S_RamFgNWuswSA.jpeg" /></figure><h3>Introduction</h3><p>With inflation rapidly rising in many countries you are faced with the harsh reality of the value of your savings going town. With bank interest rates and real wages not keeping up with inflation, you end up with less disposable income and less savings each year.</p><p><strong>It doesn’t have to be that way!</strong></p><p>There are attractive investment opportunities such as stocks, bonds, exchange-traded funds, and cryptocurrencies that carry risks but when used properly offer attractive returns to retail and institutional investors helping you secure your financial future and pave the way to your financial freedom.</p><blockquote>The choice is yours. what are you waiting for?</blockquote><h3>Introduction to Value Investing</h3><p>There are many approaches and investment strategies. The one approach that I’ll share with you is called <strong>Value Investing. </strong>Extensively covered in the excellent book called <em>“The Intelligent Investor” by Benjamin Graham</em> value investing<strong> </strong>can be defined as an investment strategy that involves picking stocks that appear to be trading for less than their intrinsic or book value. Value investors actively ferret out stocks they think the stock market is underestimating. They believe the market overreacts to good and bad news, resulting in stock price movements that do not correspond to a company’s long-term fundamentals. The overreaction offers an opportunity to profit by buying stocks at discounted prices — on sale.</p><p>This can be illustrated in the simple chart below. We have time on the horizontal axis and the asset&#39;s intrinsic value on the vertical asset. Over time the asset price fluctuates around the intrinsic value and is prone to short-term shocks e.g the CEO if the company is involved in some scandal. What often happens is that the market overreacts to such news and a large selloff occurs. <strong>The period after the selloff is a perfect buying opportunity. </strong>Over time<strong> </strong>the short-term situation changes e.g a new CEO has been appointed and the asset moves closer to its’ intrinsic price or even exceeds it. <strong>This is a great selling opportunity.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/580/1*DqswicBOqb_7Ph8p78A6Lg.png" /></figure><p>The approach was popularized by investment companies such as Warren Buffet’s Berkshire Hathaway or Orbis Investment Management because it is bringing significant returns over the long run. For example, Berkshire Hathaway’s 10-year price total return is <strong>248.9%</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/516/1*YYqRXBlVpr_dkgVVz0eAKw.jpeg" /></figure><h3>Stock market and alternative assets examples</h3><p>So far we’ve been talking about value investing only in hypothetical terms. Let’s dig into some real-world examples.</p><p><strong>Example 1:</strong></p><p>BP p.l.c. is a British multinational oil and gas company headquartered in London, England. It is one of the oil and gas “supermajors” and one of the world’s largest companies measured by revenues and profits. On 20 April 2010 off the coast of the United States in the Gulf of Mexico, a large industrial disaster occurred resulting in the largest marine oil spill in the history of the petroleum industry. This event is still referred to as the Deepwater Horizon oil spill. The impact on the company’s share price was a large selloff by investors in a relatively short period of time. After the large selloff the price of BP’s shares was oscilating between £350-£550 per share presenting an attractive buying opportunity as it was much lower than previous prices and below the company’s intrinsic value. When the world moved on and the company recovered financially period of 2018 offered an attractive selling opportunity as by then the price appreciated approximately <strong>55%</strong>. Another value investing event occurred in April 2020 as BP plunged to a loss for the first three months of the year amid rising debt and plummeting global oil prices offering an attractive buying opportunity. As of today company’s shares are <strong>133%</strong> up since that period of time presenting an attractive selling opportunity. Both of those cases are visualized in the chart below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/853/1*hSbutZcvVsC-TpQ3Ng_QnQ.jpeg" /><figcaption>BP PLC&#39;s share price over time</figcaption></figure><p><strong>Example 2: S&amp;P 500 Index ETF fund</strong></p><p>The Standard and Poor’s 500, or simply the S&amp;P 500, is a stock market index tracking the stock performance of 500 large companies listed on stock exchanges in the United States. It is one of the most commonly followed equity indices. As an investor, we are able to buy shares in an exchange-traded fund (ETF) mimicking the movement of a given index such as the S&amp;P 500.</p><p>In 2020 as a result of the panic caused by the global COVID-19 pandemic a lot of large companies were affected financially. This led to a massive 29% drop in S&amp;P 500 offering a great buying opportunity. Since then the world has moved on and in 2022 the price was up by <strong>105%</strong> offering an excellent selling opportunity. This is illustrated in the chart below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/829/1*vWP-sWFHF7f_pS1aL6FgYQ.jpeg" /><figcaption>S&amp;P 500 Index value over time</figcaption></figure><p><strong>Example 3: Bitcoin</strong></p><p>The last example is from the cryptocurrency world. Although the value investing concept was designed for shares I believe it can be applied to alternative assets as well.</p><p>Bitcoin is a decentralized digital currency that can be transferred on the peer-to-peer bitcoin network. Bitcoin transactions are verified by network nodes through cryptography and recorded in a publicly distributed ledger called a blockchain. The 1st Bitcoin bubble started forming around 2017 and it seems insignificant compared to the next bubbles happening in subsequent years. After the 1st bubble collapsed in 2018 the asset price remained in a consolidated market with lower volatility and trading volume for a prolonged period of time offering a great buying opportunity for believers in blockchain technology. Consolidation is the term for a security that is neither continuing nor reversing a larger price trend. Consolidated assets typically trade within limited price ranges until another trend emerges. The 2nd bubble started forming in 2020 offering an excellent selling opportunity with a price appreciation of more than <strong>700%</strong> in some cases if bought during the consolidated market after 2018. After the 2nd bubble collapsed around May 2021 there was another buying opportunity where a smart investor could sell some Bitcoin in the autumn of 2021 with an appreciation of more than <strong>80%</strong>. Since then the price collapsed again and recently settled at the price of around 20,000 GBP per Bitcoin offering another potential buying opportunity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/830/1*D-ATrjFex_nW9uJy1gI4SA.jpeg" /></figure><h3>Challenges with Value Investing</h3><ol><li><strong>It’s a long-term game:</strong> It may take some time for the asset below its’ intrinsic value to reach or exceed its value.</li><li><strong>Contrarian approach: </strong>As an investor, you need to be comfortable with going against the crowd. This may mean being in a negative position for some time until the asset starts appreciating again. This takes discipline as some people are uncomfortable with short-term losses and decide to sell their assets too early leading to a loss of capital.</li><li><strong>Research takes a long time: </strong>Identifying undervalued assets is very time-consuming because it involves screening a huge number of securities and it’s impossible to do it as an individual even if treating it as a full-time job. Most people have other paying jobs and don’t have time to dedicate so much time to research. The paragraph below presents an answer to this problem.</li></ol><h3>Automated Value Alerts</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/415/1*Oixb1wy_F9l5qf3Uw9OjFA.png" /></figure><p>A website called Value Alerts offers you a subscription service to notify you about the latest stock market and cryptocurrency investment opportunities. It’s based on the automated screening of thousands of tradable assets and selecting the currently undervalued ones.</p><p>The selection is based price movement, volatility, volume and financial valuation. It focuses on identifying assets that had large sellofs in a short period of time and reached price consolidation, meaning that investors significantly reduced their selling activity of the asset and the price settled on a particular narrow range for a prolonged period of time. This ensures that you won’t buy the asset in the middle of the selling spree but you can enter the trade when the investors got bored of selling. This an be illustrated by the chart below, where you can see a large selloff in end of March 2021 and price consolidation starting in April.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*2ne4wGG_LoNO0BzR.png" /><figcaption>ViacomCBS Inc. Share price</figcaption></figure><p>Once subscribed it provides you with a regular email list of assets with attractive prices. An example of a regular subscription email looks like the following:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/735/1*bv8lxLuB7vaiK9OHrhm_ww.png" /><figcaption>Example email alert from Value Alerts</figcaption></figure><p>It lists the asset name, link to a time series price chart, symbol code of an asset, and description of the asset e.g cryptocurrency, growth stock, exchange-traded funds, etc.</p><p><strong>I’m sharing it with you because it saved me a ton of time on investment research!</strong></p><p>You can use the code <strong>NEWJOINER </strong>to get a free 2 weeks trial.</p><p><strong>Link to the website: </strong><a href="http://valuealerts.mynames.website/">Value Alerts Website</a></p><h3>Conclusion</h3><p>Now that you’re equipped with an understanding of value investing and were given access to the best tool that will help you save time on researching undervalued assets, I want to wish you all the luck in your investment journey and I hope that you’ll reach your financial goals!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f91659dc5679" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>