<?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 Hajar Khizou on Medium]]></title>
        <description><![CDATA[Stories by Hajar Khizou on Medium]]></description>
        <link>https://medium.com/@hajarkhizou?source=rss-cf3521740dbd------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*JmN-m6aaaQRtPLJPP6NvOg.jpeg</url>
            <title>Stories by Hajar Khizou on Medium</title>
            <link>https://medium.com/@hajarkhizou?source=rss-cf3521740dbd------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 17:17:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@hajarkhizou/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[The Road to Credible and Value-Driven AI: Start with Asking the Right Questions.]]></title>
            <link>https://pub.towardsai.net/the-road-to-credible-and-value-driven-ai-start-with-asking-the-right-questions-560acb719312?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/560acb719312</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Wed, 20 Sep 2023 16:02:26 GMT</pubDate>
            <atom:updated>2023-09-20T16:02:26.636Z</atom:updated>
            <content:encoded><![CDATA[<h4>Your first step towards building better AI solutions.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bZeiaC6an-SCww2Z" /><figcaption>Photo by <a href="https://unsplash.com/@iammottakin?utm_source=medium&amp;utm_medium=referral">Mojahid Mottakin</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The recent advancements in generative AI have caught the attention of businesses, irrespective of their size, in implementing this technology to achieve tangible business benefits. However, many of these organizations have rapidly adopted existing AI models or embarked on developing their own without a strategic assessment of the challenges they aim to address or their readiness for AI integration.</p><p>While AI undeniably offers the potential for accelerated growth, revenue enhancement, and improved customer satisfaction, extracting its true value necessitates establishing a conducive environment for the technology to thrive.</p><p>As a CxO, it’s likely that various stakeholders, be it your board, competitors already capitalizing on AI, or third-party entities, have advocated for AI adoption in your organization. The current industry landscape doesn’t merely require an influx of AI solutions but demands credible and value-driven solutions.</p><p>Over the years of working and consulting for many organizations, I had the privilege of helping my clients design and build AI capabilities that align and drive their missions. Today, I’m sharing ‘The Road to Credible and Value-Driven AI’ with the intention of guiding the readers in their journey, bringing them closer to creating value-driven solutions and steering them away from the prevailing hype surrounding AI.</p><h3>Where do we start?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*O0gEgRZBk4yVkvMNFjscOw.png" /><figcaption>Image by Author — used <a href="https://www.openpeeps.com/">open peeps</a> Library.</figcaption></figure><p><strong>First, Always Start with ‘Why?’</strong> While it may seem intuitive, many organizations focus on the ‘What.’ Beyond merely identifying the right solution or answer, it’s imperative to <strong>ask the right questions</strong> — meaningful questions that can truly unlock value, even if they are challenging. As Simon Sinek articulates in his book:</p><blockquote>‘It all starts with Why’.</blockquote><p>Identifying the most critical issue that is addressable and offers significant potential for progress is essential. One should proceed with targeted actions only after this understanding, especially in AI implementation. Thoroughly examining the underlying cause of the issue necessitates a careful and deliberate approach. Frequently, the undertaking might be a complex and laborious endeavor, especially when confronted with intricate difficulties. Ambiguities in your information might introduce additional complexities in defining a problem. Firms must exhibit curiosity and courage, consistently asking the right questions within the right context.</p><p>In cognitive sciences, it is commonly known that there are two distinct cognitive systems underlying reasoning. System 1 is fast, focusing on readily available information, often summarized by the principle ‘What You See Is All There Is’ (WYSIATI). It excels in rapidly detecting associations, connecting the dots, and quickly developing a coherent story. As Adam Grant notes in his book ‘Originals’:</p><blockquote>“People can’t help seeing signals, even in the noise.”</blockquote><p>On the other hand, System 2 is believed to have evolved much more recently and is thought by most theorists to be uniquely human. System 2 thinking is slow and sequential. Despite its methodical nature, System 2 enables abstract and hypothetical thinking, which is beyond the capabilities of System 1.</p><p>The dynamics of these cognitive systems are prominently evident in the business landscape. Frequently, organizations oscillate between two extremes. At one end of the spectrum, there is a tendency to make rapid choices motivated by the appeal of quick action. This can result in decisions made without adequately examining the sufficiency or relevance of the available information. Such urgency to act can lead to misaligned strategies or overlooked nuances. On the contrary end of the spectrum, some organizations become excessively cautious and engage in exhaustive deliberation. They may pore through countless PowerPoint presentations, reports, and meetings to evaluate every available information. While thoroughness is commendable, this approach can sometimes result in “analysis paralysis,” the overwhelming volume of data and perspectives hinders rather than facilitates decision-making. Businesses must balance these two extremities to make informed, timely, and effective decisions.</p><p>In the corporate landscape, terms such as ‘quick win’ and ‘low hanging fruit’ frequently resonate. However, it’s not uncommon for organizations to invest significant time, often weeks or months, and substantial financial resources into tools, cloud infrastructure, and other associated costs for a ‘Proof of Concept’ (POC).</p><p>Ironically, these POCs sometimes aim to automate processes that, when assessed, would take mere minutes to execute manually. For organizations to effectively leverage the capabilities of artificial intelligence and attain a measurable return on investment (ROI), it is imperative to allocate resources prudently. Strategic discernment in identifying genuine opportunities for AI application, rather than pursuing automation for its own sake, is vital to ensure that investments yield meaningful and sustainable outcomes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*XWrl95v3p1xBqiOVTLtT5g.png" /><figcaption>Image by Author — used <a href="https://www.openpeeps.com/">open peeps</a> Library.</figcaption></figure><p><strong>Second, Keep your focus and avoid shiny object syndrome. </strong>It’s easy to be swayed by the latest innovations and trends. The allure of the ‘shiny object syndrome’ — the tendency to chase the newest technologies or trends without fully realizing their implications or alignment with business goals — can be detrimental. Businesses often invest heavily in the latest buzzwords: one moment, it’s machine learning; the next, it’s deep learning, and then the spotlight shifts to generative AI. Such frequent changes in focus can lead to significant expenditures without yielding a proportionate return on investment.</p><p>However, It is crucial to remember the identified problem and remain committed to solving it. This focus guarantees a more efficient use of resources and maximizes the possibility of releasing genuine value. While staying abreast of technological advancements is essential, ensuring that these innovations serve the organization’s core objectives and do not divert attention and resources is vital.</p><p>In addition, they will undoubtedly face uncertainties and “unknown unknowns.” These are unanticipated challenges or variables that were not initially considered but can substantially impact problem-solving. In such circumstances, organizations must allocate a buffer for experimentation, allowing adaptability and flexibility. This allows learning from unanticipated outcomes and refining strategies based on real-world feedback.</p><p>Moreover, it is essential to bring in subject matter experts in the early stages of the process. With their extensive knowledge and experience, these experts can provide invaluable insights, guide the experimentation process, and aid in navigating the complexities, ensuring the organization stays on the path to impactful solutions and value realization.</p><p><strong>Last, TAKE ACTION. </strong>Once an organization thoroughly understands the problem and has accounted for potential uncertainties, the next crucial step is to take coherent action. Coherent action signifies that every step taken is aligned with the overarching objective and consistent with understanding the problem and the experimentation phase. It’s not merely about implementing solutions but ensuring they are synergistic, building upon one another to create a holistic approach.</p><p>This requires a combination of strategic foresight, tactical execution, and continuous feedback cycles to adjust course as needed. It is also essential to create a culture of collaboration in which cross-functional teams work in partnership to drive the solution forward by leveraging their unique expertise. Organizations can optimize their efforts, reduce redundancies, and speed up the path from problem identification to solution implementation by ensuring actions are coherent and aligned.</p><p>Finally, the journey of AI implementation is as much about the question or challenge to be solved as it is about the solutions developed. The appeal of AI’s transformational promise can sometimes outweigh its practical use. However, a technology-first approach — choosing AI and then looking for challenges to solve — can result in misaligned tactics and unsatisfactory outcomes.</p><p>By asking the right questions, organizations can ensure that they are not just leveraging AI for the hype surrounding it but for its genuine capability to address specific challenges. This problem-first approach guarantees that AI acts as a tool adapted to the company’s particular needs rather than the organization bending to meet the limits of the technology. In essence, the success of AI implementation is dependent on clarity of purpose, which is obtained through asking the right questions from the start.</p><p><strong>About the Author:</strong></p><p>Hajar is an Artificial Intelligence advocate, speaker, and consultant who values technology, innovation, and excellence.</p><p>She helps mission-driven organizations leverage the power of data and AI to maximize their impact. She takes pride in listening to customers’ needs and crafting well-architected, innovative, and value-driven solutions that help her customers achieve their goals.</p><p>She can be found on <a href="https://www.linkedin.com/in/hajarkhizou/">LinkedIn</a> &amp; <a href="https://twitter.com/HajarKhizou">Twitter</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=560acb719312" width="1" height="1" alt=""><hr><p><a href="https://pub.towardsai.net/the-road-to-credible-and-value-driven-ai-start-with-asking-the-right-questions-560acb719312">The Road to Credible and Value-Driven AI: Start with Asking the Right Questions.</a> was originally published in <a href="https://pub.towardsai.net/">Towards AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Production ML: Getting Started with MLOps]]></title>
            <link>https://medium.com/data-science/production-ml-getting-started-with-mlops-4856c38b3eab?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/4856c38b3eab</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[mlops]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[editors-pick]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Thu, 17 Feb 2022 04:49:12 GMT</pubDate>
            <atom:updated>2022-02-17T04:49:12.745Z</atom:updated>
            <content:encoded><![CDATA[<h4>How Can MLOps help you productionize your ML projects? And Why do you need to start adopting MLOps ASAP?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*E5sni9fPGiQgFhpY" /><figcaption>Photo by <a href="https://unsplash.com/@markuswinkler?utm_source=medium&amp;utm_medium=referral">Markus Winkler</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Continuous Integration and Continuous Delivery (CI/CD) issues have been known in “classic” IT projects for a long time. This is why the term DevOps was born to designate the programming principles that promote the automation and monitoring of all stages of software development.</p><p>In our machine learning systems, which are increasingly beginning to be industrialized and integrated into existing IT systems, the challenge is to apply these principles to the world of data science.</p><p>Building models is hard, and putting them into production is even harder. Replicating your pipeline and results, or allowing other data scientists to do so, is equally complex. How many times have you had to abandon previous work because it was not properly documented or too difficult to replicate?</p><p>In this article, we will see how MLOps addresses this issue and how to integrate this approach into a full Machine Learning lifecycle.</p><h3>WHAT IS MLOps?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*P3uNIUmehzi_QqkphImQ2Q.png" /><figcaption>Image by Author — used <a href="https://www.openpeeps.com/">open peeps</a> Library.</figcaption></figure><p>MLOps tends to be defined as the counterpart of DevOps for machine learning. We can therefore deduce that MLOps allows the automation and monitoring of the steps of a machine learning project.</p><p>But before going further into the subject, let’s briefly recall the different stages of a machine learning project:</p><ul><li>Definition of the business need and requirements,</li><li>Data collection and analysis,</li><li>Data validation,</li><li>Data pre-processing,</li><li>Model exploration, training, selection, and validation,</li><li>Deploying a model in production.</li></ul><h3>So, where does MLOps fit into all this?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/290/1*OMgeqX3kTd8XlkTj-Ra9EA.png" /><figcaption>Image by Author — used <a href="https://www.openpeeps.com/">open peeps</a> Library.</figcaption></figure><p>Well, ideally, everywhere! But I have broken this down into three main parts that I feel are important and that I will detail below:</p><ul><li>Experimentation management,</li><li>Pipeline automation,</li><li>and model monitoring (especially functional).</li></ul><p>Not all the principles will be necessary for your projects. Instead, they are toolboxes that you can decide to apply or not, depending on your needs.</p><h3>EXPERIMENT TRACKING</h3><p>In some situations, it can become difficult to track all the experiments that have been done over time (because they are very numerous or because they were spread over many months, or because the team is very large, etc.). In addition, the code is not always kept (what is the point when we decided to try another model?). However, keeping track of these runs can be useful for auditing purposes to help debug or justify the choices made.</p><p>This is where experiment tracking comes in:</p><p>Experiment tracking focuses on the iterative model development phase when a data scientist tries different configurations to achieve the desired performance.</p><p>For example, how to avoid screenshots to historize and share project progress and quickly reproduce a previous experiment?</p><p>Experiment tracking can be summarized in a few simple principles:</p><ul><li>Concentrate and organize all experiments in one place,</li><li>Compare experiments and analyze results to debug the model,</li><li>Improve collaboration: see what everyone is doing, easily share experiment results, access experiment data (input, metrics, etc.),</li><li>Analyze and manage experiments.</li></ul><h3>And technically, how do we do it?</h3><p>Each team can use its tools to manage and follow its project’s different steps of ML experiments. However, for reasons of efficiency, it is advisable to consider specialized tools on the subject. Many open-source or software tools meet this need and can easily and securely integrate into the desired workflow.</p><p>Depending on the need, an experiment tracking tool can have different functionalities. In essence, here is a non-exhaustive list from a purely technical point of view of the services that a dedicated tool can offer us.</p><p><strong>Creation of a model:</strong> For the research and data exploration phase, it is necessary to be able to:</p><ul><li>Record, log, display the model metadata (data, metrics, tags, etc.);</li><li>Compare several executions: comparative table, graph, etc.</li><li>Create or automatically generate a dashboard/summary report;</li><li>Track hardware consumption, in terms of CPU, GPU, memory, etc.;</li><li>Be able to collaborate on the same project and share these new integrations;</li></ul><p><strong>Deployment to production:</strong> Once the data scientist selects a model, an ML Engineer must put it into production. For this, experiment tracking allows, with the help of a good tool, to :</p><ul><li>Easily and quickly reproduce and re-execute the experiments;</li><li>Track the experiments and their metadata;</li><li>Integrate the model to the automation and CI/CD chain</li></ul><h3>PIPELINE AUTOMATION</h3><p>The challenge of pipeline automation is to answer the need to automate what can be automated in the deployment of our model. After all, do we need a person to go and run a code in Jupyter Notebook every month to retrain the model in production? Wouldn’t this data scientist be more useful to create new models?</p><p>So let’s take a look at what should be put in place:</p><ul><li><strong>Version management of the code(<em>first of all of the models, but also the data and predictions)</em></strong>: for this step, it is advisable to register the code in a code management tool (Git, for example) and to set up tools managing the model registry, ML metadata store, and Feature store.</li><li><strong>Automatic retraining</strong>: the performance of a model in production degrades over time, so we must be able to retrain the model with more data or fresher data to keep up with the latest trends. Note that retraining can be triggered according to a schedule or according to triggers (e.g., performance drops below a threshold)</li><li><strong>Automatic tests</strong>: the tests must focus on the code as in a classic development cycle but also on the data.</li><li><strong>Production monitoring</strong> to be able to adapt the behavior of the model in production.</li></ul><p>If we summarize, this will give us the following diagram:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*lv368N51VQr_RhUd.jpg" /><figcaption>Credits: <a href="https://ml-ops.org/content/mlops-principles,">https://ml-ops.org/content/mlops-principles</a> (CC BY-SA)</figcaption></figure><h3>FUNCTIONAL MONITORING</h3><p>Machine learning models are dynamic and constantly evolve according to the input data. Therefore, it is safe to say that a model is at its best just before its deployment when the data scientists make the final modifications. But how do you ensure that this model remains effective over time?</p><p>Combining monitoring, logging, and alerting tools with the data processing flow prevents risks.</p><p>But what could go wrong?</p><p>The sources have changed, new data, handling errors, data normalization.</p><p>Here are the different steps we can encounter:</p><p><strong>If the data does not match your model’s expectations, it will most likely influence the performance.</strong></p><p>We can control the type, the format, the number of missing data per input variable. In addition, there is a wide range of statistical tests and tools to measure deviations in the data:</p><p>For continuous variables: divergence and distance tests such as Kullback-Leibler divergence, Kolmogorov-Smirnov statistic (widely used), population stability index (PSI), Hellinger distance, etc.</p><p>Categorical variables: chi-squared test, entropy, cardinality, or modality frequency.</p><p><strong>The relationships between the input variables and your target can also change over time.</strong></p><p>Model drifts can be checked with the same statistical tests as presented above. Also, metrics and checks should be placed to compare reality with earlier predictions. Finally, using production data (when it becomes available), we can calculate the true accuracy of our model.</p><p>A final point that can be addressed is the creation of business KPIs because we must not forget that our models are supposed to respond to business problems. Therefore, following such indicators can also help understand our model’s performance.</p><p><strong>And now that gaps have been detected, what should we do?</strong></p><ul><li>Document the problem and determine if it is a seasonal problem or a one-time aberration.</li><li>Retrain the models and apply this model to the new data.</li><li>Logging</li><li>Alerting: no need to constantly look at the dashboard; we are alerted when actions seem necessary.</li></ul><h3>CONCLUSION</h3><p>MLOps is the operationalization of Machine Learning model management. It aims to create an end-to-end process for creating, implementing, and managing repeatable, testable, and scalable machine learning models.</p><p>It is important to understand that the way to practice MLOps, by definition, is not something fixed. Processes, architectures, and tools must be adapted to the realities of your environment.</p><p>However, there are workflow models that can be used as a guide in most cases. Each of the major steps in these workflows can be executed using a specific tool. There are a number of tools, both proprietary and open-source, that facilitate the practice of MLOps.</p><p>If there is one thing to remember, it is that: MLOps is a set of practices that should be as agile as possible and should be based on the automation of the delivery and continuous integration processes of ML applications.</p><p><strong>About the Author:</strong></p><p>Hajar is an Artificial Intelligence advocate, speaker, consultant, and certified machine learning engineer who values technology, innovation, and excellence.</p><p>She helps mission-driven organizations leverage the power of data science and machine learning to maximize their impact. She takes pride in listening to customers’ needs and crafting well-architected, innovative, and scalable ML solutions that help her customers achieve their goals.</p><p>She can be found on <a href="https://www.linkedin.com/in/hajarkhizou/">LinkedIn</a> &amp; <a href="https://twitter.com/HajarKhizou">Twitter</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4856c38b3eab" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/production-ml-getting-started-with-mlops-4856c38b3eab">Production ML: Getting Started with MLOps</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Are women crushing the gender gap in Data Science and related fields?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/data-science/are-women-crushing-the-gender-gap-in-data-science-and-related-fields-3ab7ae299b4b?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/1000/1*d5VNwndC6XjMm4Oq8jb4yg.jpeg" width="1000"></a></p><p class="medium-feed-snippet">Inclusion and diversity in Academia and Industry in AI.</p><p class="medium-feed-link"><a href="https://medium.com/data-science/are-women-crushing-the-gender-gap-in-data-science-and-related-fields-3ab7ae299b4b?source=rss-cf3521740dbd------2">Continue reading on TDS Archive »</a></p></div>]]></description>
            <link>https://medium.com/data-science/are-women-crushing-the-gender-gap-in-data-science-and-related-fields-3ab7ae299b4b?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/3ab7ae299b4b</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[women]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Sat, 22 Feb 2020 03:35:30 GMT</pubDate>
            <atom:updated>2020-02-22T03:35:30.674Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Here’s How Imposter Syndrome Can END your career as a Data Scientist]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.datadriveninvestor.com/heres-how-imposter-syndrome-can-end-your-career-as-a-data-scientist-75e240f26238?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/768/1*LpCA26u9lxdsQVqjD7QDjw.png" width="768"></a></p><p class="medium-feed-snippet">And here is how to OVERCOME IT.</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/heres-how-imposter-syndrome-can-end-your-career-as-a-data-scientist-75e240f26238?source=rss-cf3521740dbd------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/heres-how-imposter-syndrome-can-end-your-career-as-a-data-scientist-75e240f26238?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/75e240f26238</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Fri, 24 Jan 2020 13:44:49 GMT</pubDate>
            <atom:updated>2020-01-24T13:44:49.717Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[15 Powerful principles to get the most VALUE of your data.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@hajarkhizou/15-powerful-principles-to-get-the-most-value-of-your-data-a8c764430026?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/1011/1*uJNhNecOCY5M96M6613-Ng.png" width="1011"></a></p><p class="medium-feed-snippet">Guidelines to produce mature Data science.</p><p class="medium-feed-link"><a href="https://medium.com/@hajarkhizou/15-powerful-principles-to-get-the-most-value-of-your-data-a8c764430026?source=rss-cf3521740dbd------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@hajarkhizou/15-powerful-principles-to-get-the-most-value-of-your-data-a8c764430026?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/a8c764430026</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Mon, 20 Jan 2020 16:28:26 GMT</pubDate>
            <atom:updated>2020-01-20T16:28:26.189Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Is Artificial Intelligence Dead?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/data-science/is-artificial-intelligence-dead-ceb3830033a1?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/2600/0*_m0S9Ri8RlekSY5o" width="4576"></a></p><p class="medium-feed-snippet">Let&#x2019;s talk stats about AI state.</p><p class="medium-feed-link"><a href="https://medium.com/data-science/is-artificial-intelligence-dead-ceb3830033a1?source=rss-cf3521740dbd------2">Continue reading on TDS Archive »</a></p></div>]]></description>
            <link>https://medium.com/data-science/is-artificial-intelligence-dead-ceb3830033a1?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/ceb3830033a1</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Sun, 19 Jan 2020 05:34:05 GMT</pubDate>
            <atom:updated>2020-01-19T05:34:05.226Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Democratizing Artificial Intelligence]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/data-science/democratizing-artificial-intelligence-42222472d3d4?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/2600/0*hbK_3hPyOZmVaHKC" width="3000"></a></p><p class="medium-feed-snippet">&#x201C;AI Black Boxes&#x201D; are for everyone.</p><p class="medium-feed-link"><a href="https://medium.com/data-science/democratizing-artificial-intelligence-42222472d3d4?source=rss-cf3521740dbd------2">Continue reading on TDS Archive »</a></p></div>]]></description>
            <link>https://medium.com/data-science/democratizing-artificial-intelligence-42222472d3d4?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/42222472d3d4</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Fri, 17 Jan 2020 18:05:32 GMT</pubDate>
            <atom:updated>2020-01-17T18:05:32.460Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How do I get started in Data Science?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/data-science/how-do-i-get-started-in-data-science-c4d38010ced?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/2600/0*CLgCK7e3gK9rLhYC" width="8192"></a></p><p class="medium-feed-snippet">Read this if you are hoping to start a data science journey by 2020.</p><p class="medium-feed-link"><a href="https://medium.com/data-science/how-do-i-get-started-in-data-science-c4d38010ced?source=rss-cf3521740dbd------2">Continue reading on TDS Archive »</a></p></div>]]></description>
            <link>https://medium.com/data-science/how-do-i-get-started-in-data-science-c4d38010ced?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/c4d38010ced</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Tue, 03 Dec 2019 16:52:12 GMT</pubDate>
            <atom:updated>2019-12-03T16:52:12.703Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Big Data From B to A: The Hadoop Distributed Filesystem — HDFS]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/data-science/big-data-from-b-to-a-the-hadoop-distributed-filesystem-hdfs-992612cbf8aa?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/2600/0*h5gWwi7WOzRl9FEw" width="4408"></a></p><p class="medium-feed-snippet">Your guide to understand HDFS concepts</p><p class="medium-feed-link"><a href="https://medium.com/data-science/big-data-from-b-to-a-the-hadoop-distributed-filesystem-hdfs-992612cbf8aa?source=rss-cf3521740dbd------2">Continue reading on TDS Archive »</a></p></div>]]></description>
            <link>https://medium.com/data-science/big-data-from-b-to-a-the-hadoop-distributed-filesystem-hdfs-992612cbf8aa?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/992612cbf8aa</guid>
            <category><![CDATA[hdfs]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[hadoop]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Mon, 25 Nov 2019 16:06:15 GMT</pubDate>
            <atom:updated>2019-11-25T16:06:15.934Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Big Data from B to A: Technology Foundations of Big Data]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@hajarkhizou/big-data-from-b-to-a-technology-foundations-of-big-data-384f877ccf49?source=rss-cf3521740dbd------2"><img src="https://cdn-images-1.medium.com/max/2600/0*oOZ-foVW6mkNYv3j" width="3771"></a></p><p class="medium-feed-snippet">The Hadoop ecosystem and beyond.</p><p class="medium-feed-link"><a href="https://medium.com/@hajarkhizou/big-data-from-b-to-a-technology-foundations-of-big-data-384f877ccf49?source=rss-cf3521740dbd------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@hajarkhizou/big-data-from-b-to-a-technology-foundations-of-big-data-384f877ccf49?source=rss-cf3521740dbd------2</link>
            <guid isPermaLink="false">https://medium.com/p/384f877ccf49</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[hadoop]]></category>
            <dc:creator><![CDATA[Hajar Khizou]]></dc:creator>
            <pubDate>Mon, 18 Nov 2019 11:22:37 GMT</pubDate>
            <atom:updated>2019-11-18T11:22:37.821Z</atom:updated>
        </item>
    </channel>
</rss>