<?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 Frantz Augustin on Medium]]></title>
        <description><![CDATA[Stories by Frantz Augustin on Medium]]></description>
        <link>https://medium.com/@AugustinCaz?source=rss-c65b0574465------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*oC9XhyX81V_5dscD28CXug.jpeg</url>
            <title>Stories by Frantz Augustin on Medium</title>
            <link>https://medium.com/@AugustinCaz?source=rss-c65b0574465------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 10:13:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@AugustinCaz/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 100 Days of Machine Learning Challenge]]></title>
            <link>https://medium.com/@AugustinCaz/my-100-days-of-machine-learning-challenge-cd637083d55d?source=rss-c65b0574465------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd637083d55d</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[100daychallenge]]></category>
            <category><![CDATA[fastai]]></category>
            <dc:creator><![CDATA[Frantz Augustin]]></dc:creator>
            <pubDate>Wed, 23 Aug 2023 16:09:39 GMT</pubDate>
            <atom:updated>2023-08-23T23:29:41.835Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SeowgpG2TsS13lI5gAoHlg.png" /></figure><p>During the men’s soccer final between France and Argentina on December 18, 2022. I had this silly idea of doing a 100-day machine learning challenge, so I can learn enough to be able to morph my face with the soccer superstar Mbappé and even change my name to Mfrantz.</p><p>Sounds fun, right? But what started as a fun challenge turned into something more serious.</p><p>How could it not? Machine learning was all over the news, trending on Twitter every day. From apps like Prisma to Midjourney, everywhere I went on social media, I would see someone turning their picture into an avatar with a new machine learning model. The hype was real and I was caught in it, except I wanted to do it on my own without third party apps.</p><p>The next day and the 85 days that follow, I spend some time building theoretical knowledge through textbooks, online courses, and YouTube videos. Some day and night, it was about 5 min of learning and some weekend I spend hours studying. I also took detailed notes on topics like overfitting, validation, ethics, and key algorithms. My favorite learning resources were:</p><p>- The 6 hour YouTube video “<a href="https://youtu.be/1vkb7BCMQd0">Making Friends with Machine Learning</a>”</p><p>- <a href="https://course.fast.ai/">Practical Deep Learning for Coders</a> free online class at fastAi</p><p>- <a href="https://a.co/d/eExchgm">Deep Learning for Coders with Fastai and PyTorch: AI Applications Without a PhD</a> (book)</p><p>I love the FastAI class — they use a top-down approach to learning, whereas most other resources does the opposite. It was a new paradigm but this will be the way I approach learning moving forward. For example, I was able to build my first machine learning app that recognizes images of birds after the first 1 hour lecture.</p><p>As for the 6 hour YouTube video by Cassie Kozyrkov, I have watched it twice and still go back to it. It’s the best explanation of machine learning without too many jargon out there in my opinion and if they are the author explain them with analogy. I know 6 hours is a bit long, but if you binge watch TV like me, it should be easy to do. There are also shorter versions of it. <a href="https://youtu.be/US2NMUfn1tQ">Here one example on Youtube</a> .</p><h3><strong>On the challenge</strong></h3><p>This was my first time doing a challenge for anything. Even though I chose 100 days, I only did 85 days non-stop from December 18 to March 13, 2023, where I practiced, read or watched something about machine learning. Then I had to stop for personal reasons, got back to it for 5 days, and had to stop again this time to play with the GPT-4 API access as well as ChatGPT code interpreter and plugin access — that will be a story for another time. I finished the challenge on Aug 15, 2023, thanks to a friend who texted me asking what book I was reading on ML. In the last 10 days, I tried a challenging project involving training an image classifier from scratch using a large dataset of about 33GB called WikiArt with over 80,000 art images.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*961Ved7Neb1xUI1OLERElg.jpeg" /><figcaption>The journey tracked with the Streaks app.</figcaption></figure><p>The idea was to build an image classifier that would identify the genre, style and artist name of a painting. I struggled with issues like lack of compute resources and memory limitations while trying to download and preprocess the data on Google Colab. After many late nights and weekend sessions iterating, I went back to the drawing board and took on a smaller but fun project where I classified images by whether they were dogs or cats, as well as their breed, using a smaller dataset. It worked — you can try it <a href="https://huggingface.co/spaces/Frantz103/dog_cat_multi_classifier">here on Hugging Face Space</a>.</p><p><a href="https://huggingface.co/spaces/Frantz103/dog_cat_multi_classifier">Dog Cat Multi Classifier - a Hugging Face Space by Frantz103</a></p><p>On thing to consider since the Oxford-IIT pets database has information on about 37 breed types, it will not be able to predict all type of dog/cat breed :)</p><p>While I didn’t fully achieve my initial goal of morphing two images from scratch, this challenge taught me a lot.</p><p><strong>Key lessons included:</strong></p><p>1. The importance of having good data</p><p>2. Always have a training and validation set — FastAI does that automatically</p><p>3. Overfitting is enemy number 1</p><p>4. Start small with the dataset and with simpler models — it will help you iterate faster</p><p>5. Don’t give up — persevere</p><p>6. Do more practice than theory — 80% practice to 20% theory is a good ratio in my opinion</p><p>7. High school math is enough</p><p><strong>Here are the tools I used:</strong></p><p>Google Colab: I love it — the free version is generous for the FastAI class. I upgraded so I could have faster machines when training a ResNet50 on the WikiArt dataset and the Pets dataset. For those unfamiliar, Google Colab is like having Microsoft Word but you can also run code in it like a notebook.</p><p>Jupiter Notebook: I mainly used this on my old 2012 Mac to learn Pandas and in the early days of the class. It’s the same as Google Colab but open source and free.</p><p>Gradio: This helped me create machine learning apps quickly to gather feedback fast. Marketing people would call that an MVP. I built two apps with it!</p><p>HuggingFace: This is where you go to find datasets, open source ML models to train on, ML apps to try, and an ML community. You can also host apps built with Gradio here for free, as well as use their no-code tools to train your own models. I describe it as LinkedIn but for ML practitioners.</p><p>FastAI: This ML library helps you write code to train models very quickly compared to other libraries. It removes complexity so you can focus on important bits like debugging and model evaluation. It’s built on top of PyTorch, a library by Meta also popular for ML work. PyTorch will be my next conquest!</p><p>Kaggle: This is where you go to participate in ML competitions. I haven’t entered one yet but will try the Titanic and dog/cat classification ones for fun.</p><h3><strong>A Little History</strong></h3><p>I first got interested in machine learning 6 years ago, mostly for digital marketing and image generation applications. For digital marketing, it was obvious — part of my work involved helping websites rank on Google’s first page, and Google used ML heavily in its content algorithms. At the time, I was spending lots of time in Google Analytics for a content marketing agency in Boston, getting very curious and wanting to make predictions. I started asking questions like could we predict which content would go viral? That led me to predictive analytics and the rest is history, with some ons and offs.</p><p>Somehow I discovered Andrew Ng’s class on deep learning via this <a href="https://www.wired.com/story/andrew-ngs-new-online-school/">wired article</a> and quote, “I am waiting for this class, fingers crossed. This is great news.” That was 6 years ago. The class was cool — with just 1 line of code in a program called Octave, Andrew wrote a program to remove background noise from audio. I completed about half of that class. I loved the theory but not the math — I couldn’t do the exercises back then.</p><p>On another note, his website <a href="https://www.deeplearning.ai/">DeepLearning.AI</a> is another great resource. They also have 1-hour classes teaching how to use different ML model APIs from OpenAI — even things on LangChain.</p><p>Later that year, I trained my first ML model recognizing handwritten digits using a Google TensorFlow tutorial on the MNIST dataset. The training worked but I had no idea what I was doing. Will I understand it? I don’t know. But, for sure, Ill keep trying and see where this deep learning adventure lead me.</p><figure><img alt="Linkedin Post from 6 years ago with my first machine learning training experiment. The post read “3 years ago, I wrote my first hello word in PHP and a few months later I was blessed with a front end developer job in Kansas. Fast forward to now, I am in Boston working for an amazing content marketing software company as a content coordinator. I love marketing but the love of code is still there. Last week, I wrote my first MMNIST, it’s like writing hello word for machine learning." src="https://cdn-images-1.medium.com/max/1024/1*cWIKSPmyv9eBhBW0vbR5AQ.jpeg" /><figcaption>Linkedin Post from 6 years ago with my first machine learning training experiment.</figcaption></figure><h3><strong>What’s Next for Me</strong></h3><p>I will keep learning about machine learning, spending more time on the math and evaluation. I’ll do more practice than theory at an 80/20 ratio, and think of ways to implement simple ML models with large language models to help with web accessibility, UX, and analytics.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd637083d55d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Do You Want To Be An Entrepreneur by  REALToDo CRM]]></title>
            <link>https://medium.com/@AugustinCaz/why-do-you-want-to-be-an-entrepreneur-by-realtodo-crm-6b9de44e7830?source=rss-c65b0574465------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b9de44e7830</guid>
            <category><![CDATA[social-entrepreneurship]]></category>
            <category><![CDATA[startup-lessons]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[eforall]]></category>
            <dc:creator><![CDATA[Frantz Augustin]]></dc:creator>
            <pubDate>Thu, 02 Nov 2017 11:01:01 GMT</pubDate>
            <atom:updated>2017-11-02T11:01:01.182Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*v4sEuum00WlgP6-NVvuHAw.jpeg" /></figure><h3>Why Do You Want To Be An Entrepreneur by REALToDo CRM</h3><p>Trying to turn your idea into a small business can be a lonely and frustrating process. But you’re not the only one making the journey — discovering the failures, successes, thoughts, and opinions of your fellow entrepreneurs can be a comfort when you feel like you’re spinning your wheels. REALToDo CRM got in touch with a couple of the EforAll entrepreneurs and asked them questions about their businesses.</p><h4>Maureen Evelina from ILoveMyNighty</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*DEceSBjh-Ewspfnz8-hu8g.jpeg" /><figcaption><a href="https://realtodo.com/entrepreneurship-for-all-at-the-makers-and-innovators-pitch-contest/">Photo by James Threadgill</a></figcaption></figure><p>Maureen Evelina describes her upcoming line of women’s clothing as “seductively snug nighties designed for lovely ladies who want to look hot when it’s cold”.</p><p>I’ve spoken with Maureen on a couple occasions — the first time was at a pitch workshop and the second was at the Makers and Innovators Pitch Contest. She’s a charismatic woman with a keen eye for design and a clear passion for fashion. After her EforAll presentation on 10/26/17, we caught up with her and asked a few questions:</p><p><strong>R: What made you want to become an entrepreneur?</strong></p><p>M: I never knew I would be an entrepreneur. I just had the idea to design for myself a certain type of loungewear that I had never seen in the market before. Then I couldn’t stop thinking about how this would be something other women would want, too and I spent a year mulling it over before I decided to see if I could turn my idea into a business- and then just went for it!</p><p><strong>R: Any advice for other aspiring entrepreneurs?</strong></p><p>M: Advice for other entrepreneurs is to seek out any and all opportunities to learn about running a business. Don’t just wing it. There are plenty of educational opportunities and people out there who are willing to help (Yay EforALL!). Being prepared is smart. Spending a few hundred or thousand dollars on classes is the best way to save money as you will learn the right way to do things and save yourself from making costly mistakes and poor decisions. People overall are very supportive and want to see you succeed. You have just as much potential to succeed as the people who started Spanx or Facebook, so don’t talk yourself out of it before you try. Do as much as you can and don’t feel like “you’re behind”. It could take 1–2 years before you are actually up and running and that’s a good thing.</p><p><strong>R: How is the business going? Any big news coming up?</strong></p><p>M: The business is going well. As per my advice above — I’m 10 months in and feel like I’ve barely touched the surface of what I need to do. It is incredibly challenging and time consuming but exciting and gratifying at the same time. I am building a strong support network and am currently in an accelerator program for fashion startups (Factory45) so I’m learning a lot of what I need to do. Having a full time job and trying to learn from scratch about business, designing, manufacturing, e-commerce, branding, marketing, social networking — all at once — has my head spinning, but I’m devoted and want to see all this become a reality and enjoy my success in the next couple of years. Up next — I was selected as a graduate of the business planning class through the Center For Women in Enterprise to be in another pitch contest in two weeks at LISC Boston. I find out today if I’m accepted, so I will give it another shot if the opportunity is there!</p><p><strong>R: What did you enjoy most about the EforAll event?</strong></p><p>M: What I enjoyed most about the EforAll event was being a part of something so supportive. Everyone there was kind and excited. I have never done anything like it before and didn’t really know anything about pitch contests, so I was glad that I went outside of my comfort zone to get my name out there and get good advice in return.</p><p><strong>R: How can users get in touch with you, follow you, etc.?</strong></p><p>M: I would love for people to sign up for my newsletter on <a href="https://ilovemynighty.com/">ILoveMyNighty.com</a> so they can stay up to date on my progress and my upcoming launch in 2018.</p><p><em>Find out more about Maureen and her business here:</em></p><p><em>Website: </em><a href="https://ilovemynighty.com/"><em>https://ilovemynighty.co</em></a><em>m</em></p><p><em>Facebook: </em><a href="https://www.facebook.com/ilovemynighty"><em>https://www.facebook.com/ilovemynighty</em></a></p><p><em>Twitter: </em><a href="https://twitter.com/ilovemynighty"><em>https://twitter.com/ilovemynighty</em></a></p><h4>Frantz Augustin from Clubartizan</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*q0gXNjZq9DYzJJFA.jpg" /><figcaption><a href="http://frantzaugustin.com/">Photo by Frantz Augustin</a></figcaption></figure><p>Clubartizan is described as a “social impact marketplace for high quality arts and crafts from artists around the world”.</p><p>I met Frantz Augustin for the first time at the Makers and Innovators Pitch Contest. During our short interaction, he proved himself to be an accomplished and knowledgeable programmer, as well as a talented photographer. He also graciously agreed to answer our questions after the EforAll event:</p><p><strong>R: What made you want to become an entrepreneur?</strong></p><p>F: My high school principal is the reason why I want to become an entrepreneur. He was always talking about one problem that needed to be solved: extreme poverty. In my country, 59% of the population live under $2.41 a day. As a native from there, I want to help and one way to do this is by creating jobs for the poor.</p><p><strong>R: Any advice for other aspiring entrepreneurs?</strong></p><p>F: First, the journey is hard, but do not let that stop you. If you feel the calling, go for it. However, It’s not going to be easy. For instance, I am not sure if Clubartizan will work, but I am happy trying it and I’ll give it my all.<br>Lastly, learn from other entrepreneurs. My go to entrepreneurship article is “ How to Start a Startup by Paul Graham”. My favorite entrepreneurship books are “The Hard Thing About Hard Things by Ben Horowitz”, “Ecosystème by Rachel Vanier”, “Chaos Monkeys by Antonio García Martínez”, “How Google Works by Alan Eagle, Eric Schmidt, and Jonathan Rosenberg” and “Give Work: Reversing Poverty One Job at a Time by Leila Janah”.</p><p><strong>R: How is the business going? Any big news coming up?</strong></p><p>F: I am at the early stage right now. I had the idea 5 years ago but I was not fully invested in it. I had to learn how to code, graduate from college and get experience in the real world. Right now, I am all in and I am very excited about it. I have been applying for a lot of accelerators and I am working on finishing the prototype. With or without accelerators, I will launch Clubartizan.</p><p><strong>R: What did you enjoy most about the EforAll event?</strong></p><p>F: It was a great event. The last time I went to an event like that was when I was in college. I enjoy meeting other entrepreneurs and seeing what they are working on. I also appreciate the feedback I received. The event was a great source of motivation for me as well.</p><p><strong>R: How can users get in touch with you, follow you, etc.?</strong></p><p>F: My twitter and instagram handle is @augustincaz. I am also on LinkedIn at <a href="https://www.linkedin.com/in/frantz103/">https://www.linkedin.com/in/frantz103/</a>. My email is <a href="/service/mailto:hirefrantz@gmail.com">hirefrantz@gmail.com</a>.</p><p><strong>R: Anything else about your company that you would like to highlight?</strong></p><p>F: If you meet art lovers or artists send them to <a href="https://clubartizan.com/">clubartizan.com</a>. The access code is dream. If you are an artist or an art lover, please reach out to me — it’s the right thing to do</p><h3>Launch your ideas with EforAll</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*I9DAwO2RHXLxzolP.jpg" /><figcaption>Photo by Taylor Tofal / <a href="https://realtodo.com/author/taylor/">https://realtodo.com/author/taylor/</a></figcaption></figure><p>EforAll provides a great service for entrepreneurs who need some help getting traction with their target markets. UbiquiTools presented <a href="https://realtodo.com/">REALToDo CRM</a> at an EforAll event in Lowell on 10/12/17 — and won 3rd place! More important than the prize money, though, was getting to speak with local business owners. EforAll staff and judges asked tough (but necessary) questions and were generous with feedback and constructive criticism.</p><p>If you’re looking to get serious about your business, take a look at EforAll’s <a href="https://eforall.org/pitch-contests/">upcoming pitch contests</a> and <a href="https://eforall.org/business-accelerator/">business accelerator</a>.</p><blockquote>Originally published by James Threadgill at <a href="https://realtodo.com/entrepreneurship-for-all-at-the-makers-and-innovators-pitch-contest/">https://realtodo.com/entrepreneurship-for-all-at-the-makers-and-innovators-pitch-contest/</a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6b9de44e7830" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>