<?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 James Sohn on Medium]]></title>
        <description><![CDATA[Stories by James Sohn on Medium]]></description>
        <link>https://medium.com/@sohn21c?source=rss-3b7e5c97c3d0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*wLHloDl5UpbzmJHRKRuP_w.jpeg</url>
            <title>Stories by James Sohn on Medium</title>
            <link>https://medium.com/@sohn21c?source=rss-3b7e5c97c3d0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 00:49:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sohn21c/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[My data science internship as robotics student]]></title>
            <link>https://tech.cars.com/my-data-science-internship-as-robotics-student-f480d4a0792f?source=rss-3b7e5c97c3d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/f480d4a0792f</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[internships]]></category>
            <dc:creator><![CDATA[James Sohn]]></dc:creator>
            <pubDate>Mon, 04 Nov 2019 15:23:40 GMT</pubDate>
            <atom:updated>2019-11-04T17:24:11.360Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Who Am I ?</strong></h3><p>Where do we start? I guess I should start by giving you the sight over the forest before getting to the trees. I am a graduate student pursuing a Master’s degree in Robotics. I am a mechanical engineer by training and have 3+ years of automotive engineering experience before coming back to school. And I chose to come work for Cars.com during the summer as a data science intern. Still interested in learning more about my experience? Let’s get one step closer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/769/1*WMUQM7RFLh3-ej0_M1hHMQ.jpeg" /><figcaption>[Google image](<a href="https://www.sierraclub.org/sierra/forest-conservation-part-climate-conversation-too">https://www.sierraclub.org/sierra/forest-conservation-part-climate-conversation-too</a>)</figcaption></figure><h3><strong>Why Data Science?</strong></h3><p>You may wonder why I made such a choice to do a data science internship with my background in Robotics or Mechanical Engineering, even. I decided to come back to school because I wanted to learn more about the computer vision. While immersing myself into the world of buzz words, like deep learning, machine vision and artificial intelligence to educate myself in the field of computer vision, I, without knowing, started becoming curious about the data science. After all, images and videos are not the only form of data I can build the machine learning model with.</p><p>What makes the data science, and machine learning in general, particularly compelling to me is that humans are giving computers novel capability they never possessed before, prediction based on learning. Might I add to it, the ability to predict the outcome(or an object in object detection task) based on previous learning. Sure, we’ve used the computers to develop a model that extrapolates the trend analysis to predict what will happen next, but it was still bound by the limitation of such model and the environment it was computed within. It never had ability to see many pictures of an apple in a box and then recognize a different apple on the table. You see the difference? Machine learning allows learning what is relevant from the given information by training and making predictions to best match its understanding that may reside outside the context it was trained in the first place. The prediction of such machines won’t always be right as we, humans, are not always right, but what it can do now sounds a lot like what a human can do.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UsYsbsZd-CE3CTkqrwRIwg.jpeg" /><figcaption>[Datanami](<a href="https://www.datanami.com/2018/09/17/improving-your-odds-with-data-science-hiring/">https://www.datanami.com/2018/09/17/improving-your-odds-with-data-science-hiring/</a>)</figcaption></figure><h3>Data Science at Cars.com</h3><p>Data science at Cars.com plays a significant role. As discussed earlier, the machine learning techniques used in data science allows for prediction. To name a few examples of use cases, it predicts what a consumer would like to see next on a web page — recommendation. It predicts if the listed vehicle will be sold soon — “Hot Car” badging. It even provides select customer reviews that are deemed relevant for each individual car shopper for specific models of a vehicle. These tools provide the consumers searching on the Cars.com platform a seamless experience while they shop for their perfect car.</p><p>Data science leadership at Cars has created a great community of data scientists where you can learn, teach and collaborate with fellow team members. Such an environment is critical in that you are able to openly discuss what you are struggling with and get help from other team members, which I will discuss more in detail below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*lahkWhB7s3fAGEbDD0DY8A.png" /><figcaption>[Cars.com](https://www.cars.com)</figcaption></figure><h3>During my internship at Cars.com</h3><p>During the 10-week internship, I was tasked to work on a project to predict the quality of a lead a consumer submits to a dealership on the Cars.com platform leveraging the machine learning technique. To elaborate, a consumer goes on Cars.com and might perform different combinations of actions. Search, click, browse or even read an article in the Research section. He or she, after a chain of such actions, may submit a request for more information or a “lead” to the dealership with interest in purchasing a specific vehicle on that dealership’s lot. The objective of the model we tried to build was to predict the possibility of conversion of the submitted lead to the sale of a vehicle. My job was to engineer the model previously developed by the senior member of data science team, that was to be evaluated with AUC score and automate the process of adding more training data in the future.<br> <br> It was a great challenge for me in many different aspects. Not only have I not dealt with code base in such scale but also there were machine learning techniques used in the model that were somewhat novel to me. I spent a lot of time reading before adding new lines to it, trying to understand how pre-processing of the data was done and how features were created. In the process of all this, I had many sessions with both my supervisor and the director of the team to discuss relevant machine learning algorithms and the models that were extremely helpful for me to better understand the pipeline. Fast forward many weeks, the data collected in previous weeks were newly introduced to the model and the model was trained many times. Although marginal, performance of the model surely improved and the feature, automating the process of adding more training data, was successfully added to the model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xO-Eg4kRrazAZrE669KI2Q.gif" /><figcaption>[Veterans Affairs](<a href="https://www.blogs.va.gov/VAntage/252/ok-back-to-work/">https://www.blogs.va.gov/VAntage/252/ok-back-to-work/</a>)</figcaption></figure><h3>What did I learn from my internship?</h3><p>As a student exploring the field of data science and applied ML, working directly on a productionized model through the entire pipeline from fetching the data from the data base to making predictions was an invaluable experience. Students, often in the academic setting, may easily have tunnel vision and only focus on the modeling part of the machine learning. It is tempting to overlook the importance of thoroughly understanding the data, creating the relevant features out of the available dataset after proper processing and critically evaluating the performance of the model within the context of business. And the milestones of the project I was tasked with during my internship were carefully planned so that I left with the big picture of how the data science model is planned, developed and implemented in a professional environment.</p><p>Looking back at the 10 weeks at Cars, I feel blessed to have established both professional and personal connections with great data scientists in the office. I feel more equipped and capable in the realm of machine learning and data science. The internship allowed me to clearly understand my strengths and weaknesses and I came back to the campus excited and ready to get back to work. I am very much grateful for all the support I was provided with by my supervisor and the rest of the team members I was fortunate to share the office with. I would highly recommend being part of this great community of data scientists if you are considering an opportunity at Cars.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f480d4a0792f" width="1" height="1" alt=""><hr><p><a href="https://tech.cars.com/my-data-science-internship-as-robotics-student-f480d4a0792f">My data science internship as robotics student</a> was originally published in <a href="https://tech.cars.com/">Cars Commerce Technology</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>