<?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 Daniiar Abdiev on Medium]]></title>
        <description><![CDATA[Stories by Daniiar Abdiev on Medium]]></description>
        <link>https://medium.com/@daniiar.abdiev?source=rss-a8e9a15e0e51------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*FC1g_8m0-rlP04jJ.</url>
            <title>Stories by Daniiar Abdiev on Medium</title>
            <link>https://medium.com/@daniiar.abdiev?source=rss-a8e9a15e0e51------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 07:10:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@daniiar.abdiev/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[Why We Need an AI Coding Assistant Designed for Data Scientists]]></title>
            <link>https://medium.com/better-programming/why-we-need-an-ai-coding-assistant-designed-for-data-scientists-b2eff2bc090c?source=rss-a8e9a15e0e51------2</link>
            <guid isPermaLink="false">https://medium.com/p/b2eff2bc090c</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Daniiar Abdiev]]></dc:creator>
            <pubDate>Fri, 07 Oct 2022 16:51:21 GMT</pubDate>
            <atom:updated>2022-11-25T08:30:58.164Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Copilot is great but not enough for data science, it does not have proper Notebook support, SQL is useless, no data awareness, and cell output is not taken into account. That’s why we decided to build CodeSquire.ai — an AI code-writing assistant designed specifically for data professionals.</em></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*teiZjOF_7B0RXILIuYgOpA.png" /><figcaption><a href="https://codesquire.ai/">Source</a></figcaption></figure><p>Data scientists are one of the most in-demand jobs today. Companies are looking for people who can help them make sense of the vast amounts of data they have. And, as more and more data is generated, the demand for work involving data science is only going to increase.</p><p>Because of the ever-increasing need to process data, thousands of companies are looking for ways to streamline the work of data scientists, from data extraction companies like Stitch and Fivetran to MLOps companies like Weights&amp;Biases.</p><p>One of the biggest parts of data science is simply writing code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jOUD9orYfxUTAbA4jedAPA.png" /></figure><p>We are constantly writing code to clean, transform, visualize, and model data. Naturally, it would be great if we could have a code-writing assistant that would help us write code.</p><p>There are certainly great AI code-writing assistants out there like Github’s Copilot. But they are not tailored to the needs of data scientists. We need an AI code-writing assistant that understands the domain of data science.</p><p>So what features an AI code-writing assistant for data scientists should have?</p><h3>1. Support for Jupyter Lab, Jupyter Notebook, Google Colab, Databricks Notebooks, and R studio.</h3><ul><li><strong>Jupyter Notebook and Jupyter Lab</strong> — As of this moment, Github Copilot supports only VS Code, JetBrains, and Atom IDEs. While certainly lots of data scientists use VS Code and Pycharm, even more, data scientists use Jupyter Notebook and Jupyter Lab in everyday work. The notebook interface is perfect for data exploration and iteration on data problems.</li><li><strong>DataBricks Notebooks </strong>— Databricks ecosystem is growing very quickly. More and more companies are turning to Databricks for data lakehouse solutions and more and more ML engineers use DataBricks’s MLFlow for their MLOps operations. As more businesses adopt DataBricks, the number of users of Databricks Notebooks (Jupyter-like notebooks running on data bricks clusters) increases. I personally interviewed several people who have expressed interest in CodeSquire.ai’s support for Databricks Notebooks.</li><li><strong>Google Colab</strong> — I love Google Colab! Every time some new cool model drops in HuggingFace or in PapersWithCode, I try to quickly start it in Google Colab. Often this quick start turns into a full-on project. Support for Google Colab is essential. Too many data scientists love those free GPUs.</li><li><strong>R studio</strong> — R is a very popular language in data science. In fact, it is in second place right after Python. So supporting R studio, the most popular IDE for R is a must.</li></ul><p>CodeSquire.ai is already available in Google Colab and will be launched on</p><p>VS Code, Jupyter Notebook, and JupyterLab in three weeks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i-nH9V6yFhFDNf6A2qQKuQ.png" /></figure><h3>2. Notebook Cell Outputs and Markdown</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2UokkZK7pa_VAPTpVIq2fg.png" /></figure><p>As can be seen from the screenshot above, Github Copilot does not take into account cell outputs when it generates the code. And while in the example above the fix would be fairly straightforward, one can easily see how in the real project the notebook cell outputs are crucial to take into consideration.</p><p>Data Science work is an iterative process with lots of small experiments. Notebook cell outputs often guide our decision-making on what to look into next and what to explore. So it is extremely important to provide notebook cell output context to the AI to guide it to a better more precise output.</p><p>Moreover as can be seen from the screenshot above current AI code-writing assistants do not take markdown into account.</p><p>Data scientists, use markdowns a lot. To separate the data processing into several logical steps and to let people who might review the notebook know what they are looking at. It is a great way to help viewers of your notebook understand the code.</p><p>That’s why AI coding assistant for data science should take markdown into account when generating code suggestions.</p><h3><strong>3. SQL support that takes into account multiple tables</strong></h3><p>As of now SQL code generation with Copilot is mostly useless. If you want to get a relevant output you need to provide an entire table schema, and most queries usually require more than one table. This means specifying the dataset schema of several tables in comments just to get the right generation, which in turn can take as much time as writing SQL query from scratch.</p><blockquote>The ideal AI coding assistant for data scientists would make proper SQL querying support a priority.</blockquote><h3><strong>4. Translation between libraries and languages for optimization purposes. (From Pandas to Spark)</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*ob5eUKQJ0ZZ6-ejXdFAlww.png" /></figure><p>Data Scientists often use pandas. Pandas is great but not the best fit for big data ETL operations. So SQL, Dask, Modin, or Spark are better alternatives for big data production loads. Data Scientists often find themselves converting Pandas to SQL, or searching for Dask or Spark methods similar to pandas.</p><blockquote>A feature that would allow data scientists to translate Pandas to more performant alternatives with a click of a button would be very helpful.</blockquote><h3><strong>5. Data exploration Recommendations</strong></h3><p>Every data science project requires lots of data exploration.</p><blockquote>So having an AI assistant that can not only write code but also recommend the next steps would be amazing!</blockquote><p>For example, after calculating the mean of some columns, AI might recommend looking into the median and box plot to check for outliers. If you are making a pipeline to detect sentiment, the AI assistant might recommend a state-of-the-art algorithm that you have not heard of yet, or the opposite can recommend a less accurate but more efficient algorithm depending on your use case.</p><h3>Conclusion</h3><p>Creating an AI assistant that would satisfy the features above, requires a Data Science domain focus. That’s why at CodeSquire.ai we decided to build an AI coding assistant specifically designed for data professionals from the start.</p><p>All the ideas and more are being developed at CodeSquire.ai</p><p>So if you want to try an AI coding assistant developed for data scientists by data scientists, you can start today by signing up using this <a href="https://codesquire.ai/?utm_source=medium&amp;utm_medium=article&amp;utm_campaign=why">link</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b2eff2bc090c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/better-programming/why-we-need-an-ai-coding-assistant-designed-for-data-scientists-b2eff2bc090c">Why We Need an AI Coding Assistant Designed for Data Scientists</a> was originally published in <a href="https://betterprogramming.pub/">Better Programming</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>