<?xml version="1.0" encoding="utf-8"?><feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en"><generator uri="https://jekyllrb.com/" version="3.10.0">Jekyll</generator><link href="https://marcodifrancesco.com/feed.xml" rel="self" type="application/atom+xml" /><link href="https://marcodifrancesco.com/" rel="alternate" type="text/html" hreflang="en" /><updated>2026-08-15T10:33:12+00:00</updated><id>https://marcodifrancesco.com/feed.xml</id><title type="html">Marco Di Francesco Portfolio</title><subtitle>📊 Help extracting Meaning out of Data | 🇪🇺 EIT Data Science Msc | 🙋‍♂️ 2024 EIT Student Ambassador</subtitle><entry><title type="html">Retrospective - Future Work</title><link href="https://marcodifrancesco.com/2026-08-03-retrospective-future-work/" rel="alternate" type="text/html" title="Retrospective - Future Work" /><published>2026-08-03T00:00:00+00:00</published><updated>2026-08-03T00:00:00+00:00</updated><id>https://marcodifrancesco.com/retrospective-future-work</id><content type="html" xml:base="https://marcodifrancesco.com/2026-08-03-retrospective-future-work/"><![CDATA[<p>In Apr 2024 in the Netherlands I wrote the artcle “Future Job” (in Notion). Here I mentioned my wish-job/wish-life list with a 2 years outlook, and now the time has come.</p>

<h2 id="trigger">Trigger</h2>

<p>Before the review, this article started with me desperetly looking for an answer which frustrated me:</p>

<p><strong>What did I learn in the last year?</strong> This is serious. I can’t come up with an answer. Excluding the first 3 months at work where you learn new technologies and methods to work, afterwards it’s: nothing. I mean, now my whole job is doing small data model tickets, and the only real project about ML lasted 3 weeks, and my boss was so dissatisfied by it at the level he did the whole thing from scratch.</p>

<p><strong>Luca</strong> I asked this question to Luca today and the answer surprised me. He told me that in his work experience he had learned indeed: nothing. I mean, in his past has dropped a PhD, and at the age of 35 decided to drop the job and make a Master’s degree in Data Analysis, I guess that he knows what studying means. But then, if he is saying that he has learned nothing, why are people happy in some jobs, and less in others? What does make you say “this was a great day”? (Except from fridays spent talking about girls and life). To this I still don’t have an answer. But this led me in thiking: well, I can just check what I wanted a while ago and review it. If I succeeded I should consider myself happy no?</p>

<h2 id="review">Review</h2>

<p>Two years ago about in this period, I wrote my mid and long term goals, and here they are:</p>

<blockquote>
  <p><strong>Outside</strong> the workplace do fixed activities 2/3 evenings every week, e.g. going in a bar with friends or sports.</p>
</blockquote>

<p>✅ Yes, definitely. I go out with Rugg, Wanda, at least once during the week, and make a trip or go out on the weekend. This is true, definitely, but it’s not about work. This article is about my work expectations and not how much I succeded in life (I guess, assuming they are not related, I won’t get deeper).</p>

<blockquote>
  <p><strong>ML engineer</strong> to get a good understanding of the Technical challenges of a ML project.</p>
</blockquote>

<p>🔀 It’s probably a no. I have worked only 1 month FTE in ML, and the result wasn’t even published. The fact is that I can still sell it in the CV, so I will mark it as a kind-of.</p>

<blockquote>
  <p><strong>30% programming</strong>, 40% idea creation, 30% make presentation and presenting</p>
</blockquote>

<blockquote>
  <p>I have to be able to be in an environment where I can improve my ability of speak and explain myself.</p>
</blockquote>

<p>🔀 Kind-of. Most of the time I’m not programming and rather thinking about solutions, thus would make this correct, but I am also not really presenting anything. Antonio is basically doing the talking.</p>

<blockquote>
  <p><strong>Field</strong>: I should be working in a field of my interest. Best would be Finance, otherwise something I have familiarity with. Avoid companies which I don’t like to study more in depth the topics under the machine learning models, e.g. biomedicine.</p>
</blockquote>

<p>✅ Shit if this is correct. I am seriously considering how much I should be happy about the place and the field I’m in. I recall how much I didn’t give a shit about finance and how much I am into learning topics like <em>riscatti retroattivi</em> now. I should consider myself lucky in this one.</p>

<blockquote>
  <p>I should receiving the requirements a product manager and I take decisions on what is the best process to follow, explain it to my manager, and give further ideas on how to make improvement.</p>
</blockquote>

<p>✅ Yes. I am getting the business requirements and implementing it myself. I can’t complain even about that.</p>

<blockquote>
  <p>How Sept 2026 will look like: I have gained the reputation of being a knowledgable ML engineer with a good ability to explain concepts. People come to me to ask for feedback and I’m often talking about other’s tasks even though I’m not involved in the project.</p>
</blockquote>

<p>🔀 Ok fine, it’s not ML engineering, but I can’t even say I failed it. I am constantly being asked about my opining on how to do a task by Antonio. I know that most of what he asks me I find it very basic and boring, but yet he asks me and I have knowledge that is meaningful even if it’s about knowing which directories have SQL files in our disk.</p>

<h2 id="thoughts">Thoughts</h2>

<p>I am aware I am complaining and feeling down these days, but I should also appriciate how lucky I am. The first job offer I got was: a company in Finance, and in Switzerland. I am passionate as hell about finance, and even watching in my free time videos from Ben Felix, explaining financial topics most people don’t give a shit about. Having a job in a field I’m not intereted into leads me to feel like Ruggero now in the bank, don’t forget about how much I didn’t care about the <em>rene policistico</em>.</p>

<p>I have no idea why this article is not making me feel better. It should make me feel like I found out that I reachd my dream field, having good colleagues with which I am going out and having fun, while being in Switzerland and making a lot of money close to my sister.</p>

<p>It’s crazy to think that even in these conditions I am feeling sad.</p>

<h2 id="i-want-more">I want more</h2>

<p>I know, I know. You are happy, and yet you want more. This is insane to write. But let’s thinking it in a different lens. Getting stuck at this age in a satisfactory job is not good in my age. Let’s jot down a big point for my next job, something I don’t have in my current role.</p>

<p><strong>Whatever the company is, how much they pay you, and regardless of the location, I should have the feeling I’m becoming senior in my position, seniority is marked by knowledge, and transferable in a job position in another company I would like to have.</strong></p>

<p>I know, this is a broad statement. No KPI, no numbers. This is to say that the biggest cons of this job, as also stated by Gabriele, is that you learn nothing new. Having no senior person in the team means getting no guidance and doing things on my own.</p>

<p>Let’s make an example, I have currently being studying k8s for 10 months. Getting a job which uses even basic Kubernetes would make me more senior in that, and I would be using a technology I want to persue even deeper in the following job.</p>

<h2 id="conclusion">Conclusion</h2>

<p>I know that this article is kind of empty. I do not have a solution at the time of writing, I only have disconnected thoughts and much more to figure out.</p>

<p>The only thing I want to take out of this article is that I love the fact that I am in Switzerland, and I am in Finance, I am lucky to be here and when looking for a following job search for a place where I can earn knowledge in something I want to persue in the future. That’s it.</p>]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[In Apr 2024 in the Netherlands I wrote the artcle “Future Job” (in Notion). Here I mentioned my wish-job/wish-life list with a 2 years outlook, and now the time has come.]]></summary></entry><entry><title type="html">Credit Card Switzerland - Cashback</title><link href="https://marcodifrancesco.com/2025-12-28-credit-card-switzerland-cashback/" rel="alternate" type="text/html" title="Credit Card Switzerland - Cashback" /><published>2025-12-28T00:00:00+00:00</published><updated>2025-12-28T00:00:00+00:00</updated><id>https://marcodifrancesco.com/credit-card-switzerland-cashback</id><content type="html" xml:base="https://marcodifrancesco.com/2025-12-28-credit-card-switzerland-cashback/"><![CDATA[<p>Yeeee another let’s-optimize-the-small-cents rabbit hole. Credit card now.</p>

<p>After hearing it once by Emanuele, a colleague, saying “sheeeet I could have saved 2% if I purchased the forniture for my new house with American Express, over 3k that’s quite a bit”. My cheap AF brain went in the mode “<em>shit I’m missing a huge opportunity, I must spend 5-10 hours to understand what’s the best credit card and exploit this huuuuge opportunity</em>”.</p>

<p>But this was something like six months ago, I fortunately waited a week for the dopamine effect to fade. I processed the fact that American Express (Amex from now) is nowhere close to the acceptance rate of Mastercard and Visa, and it’s just not worth going shops and seeing transactions fail in front of a tired-as-f Aldi cashiers.</p>

<p>This is until Daniele and Riccardo, my ex-flatemates, came up with a screenshot of a 600 euros purchase with a whopping 6 euros cashback on a Mastercard. And here we go again with another dopamine hit. I’m a serious addict.</p>

<h2 id="now">Now</h2>

<p>So let’s see my current status. I’m now using Revolut as only <em>card account</em> I use in shops and online. Their multi-currency card, together with the perfect app has never failed me. When paying in Italy and Switzerland I don’t have to worry about forex commissions, and everything else about their service is free. The perfection until now.</p>

<p>I even went so far to purchase their gray beautiful card. [Please, I’m aware, it’s a worthless piece of plastic. Thank you for letting me notice. But it’s beautiful. Stop judging.]</p>

<p><img src="/assets/img/2025-12-28-credit-card-switzerland-cashback/revolut-shiny-card.jpg" alt="revolut-shiny-card.jpg" width="400" class="center" /></p>

<p>The point here is that Revolut has no cashback for basic accounts, and this is typical for Swiss banks. In this country it seems like the cashback is reserved for credit cards only, meaning that there is no debit card that offers cashback on no-maintence-fee accounts.</p>

<p>p.s. before you ask, Trade Republic, the large digital bank that competes with Revolut is currently not available for residents of Switzerland.</p>

<h3 id="credit-card">Credit card</h3>

<p>Gut gut. So now let’s compare the December 2025 credit card options. These are all the credit cards which are offered at Swiss residents, comparing the cashback for each one. Again, these options are restricted to no-management-fee accounts only.</p>

<table>
  <thead>
    <tr>
      <th>Card</th>
      <th>Circuit</th>
      <th>Cashback</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td><strong>American Express</strong></td>
      <td>NA*</td>
      <td>NA*</td>
    </tr>
    <tr>
      <td><strong>Certo One</strong></td>
      <td>MC</td>
      <td>1% on 3 favorite merchants • 0.25% on the rest</td>
    </tr>
    <tr>
      <td><strong>Swisscard Cashback</strong></td>
      <td>Visa or MC + Amex (mandatory)</td>
      <td>1% on Amex • 0.25% on Visa / MC</td>
    </tr>
    <tr>
      <td><strong>Poinz</strong></td>
      <td>Visa + Amex (mandatory)</td>
      <td>1% on Amex • 0.25% on Visa</td>
    </tr>
    <tr>
      <td><strong>Migros Cumulus</strong></td>
      <td>Visa</td>
      <td>0.33% in Migros credits</td>
    </tr>
    <tr>
      <td><strong>Coop Supercard</strong></td>
      <td>MC</td>
      <td>0.33% in Coop credits</td>
    </tr>
  </tbody>
</table>

<p><em>*American Express as circuit has no direct option for cashback cards. You have to pass through a card provider.</em></p>

<p>All the companies are converging to converge on the same numbers. They have cashback of 1% for Amex, 0.25% for Visa / Mastercard, and 0.33% for grocery stores.</p>

<p>As mentioned before, Amex is not for me. I don’t want to check each store if it is accepted. It’s not worth the effort. → My choice is a Visa or Mastercard which is accepted in the whole world, with a little cashback.</p>

<h3 id="impact-on-me">Impact on me</h3>

<p>Good, but now let’s see the impact it would have on me monthly. I assume a best case scenario of 1% cashback on all transactions, including physical like groceries, and digital ones like trains and Amazon. Since in Revolut I have a monthly average april to november 2025 of 600 chf-equivalent per month, the 1% would accrue to a whopping 6 CHF.</p>

<p>Let’s put this in perspective:</p>

<ul>
  <li>
    <p><strong>Physical goods</strong>. This amount of money per month, would not even buy a full sandwitch in the cheapest bar of Milano centrale central station. [Byw, kind of nice sandwiches. Approved by a starting Marco in endless trips back and forth in Chioggia.]</p>

    <p><img src="/assets/img/2025-12-28-credit-card-switzerland-cashback/bistrot-centrale.jpg" alt="bistrot-centrale.jpg" width="300" class="center" /></p>
  </li>
  <li>
    <p><strong>Time</strong>. If I’d consider the additional hours spend for yearly accounting costs, at my current net hourly rate, I probably wouldn’t inside the cashback returns.</p>
  </li>
</ul>

<h3 id="chf-only">CHF only</h3>

<p>Yeeeeeaaah, I might have written until here forgetting a huge detail. From a Reddit comment:</p>

<blockquote>
  <p><em>I would advise you to NEVER use regular credit card like Certo or Poinz outside Switzerland. They have an insane bad currency exchange rate. Instead use Neon or Revolut</em></p>
</blockquote>

<p>Searching a bit more carefully, there is no equivalent to multi-currency credit cards like Revolut, since in Switzerland credit cards are CHF-billed. Gemini and gpt both strongly agree on this.</p>

<p>So, this additionally means:</p>

<ul>
  <li>I would need to think each time I make a transaction if it’s in CHF or EUR and use the appropriate card, and</li>
  <li>My considerations about savings should be adjusted to around 300 CHF since I would’t be using use the card in Italy :(</li>
</ul>

<h3 id="conclusions">Conclusions</h3>

<p>In conclusion, the cashback alone is not a good reason to have a credit card. Under these conditions I would worry about “Am I paying in CHF, or EUR? Should I use the credit card or Revolut?”. Just not really my thing to save ~1 CHF per month.</p>

<p>The only benefit that’s coming up to mind is to have a separate credit card for car rental deposits. Buuuuuut I’m tired of writing. Ye, I know, it’s a few sentences, but still took 3+3+something hours somehow. I just want to publish. The car rental deposit is another analiziz paraliziz spiral I will get in, in the future.</p>

<p>Thanks. Bye.</p>

<p>P.S. A wise me a few days ago:</p>

<blockquote>
  <p><em><strong>CHFC first</strong>. One hugh point to not forget is the the 2.5% yoy expected return I have on CHFC (standard five percent yoy, then split in two). Always favor long term gains compared to the short term savings.</em></p>
</blockquote>

<p>Shit if I was wise to calculate that.</p>

<p>P.S.S. Nah, for now I’m not explaining what it means. I might in the future. Bye for real now 🧡.</p>]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[Yeeee another let’s-optimize-the-small-cents rabbit hole. Credit card now.]]></summary></entry><entry><title type="html">Non Zürich?</title><link href="https://marcodifrancesco.com/2025-10-05-non-zurich/" rel="alternate" type="text/html" title="Non Zürich?" /><published>2025-10-05T00:00:00+00:00</published><updated>2025-10-05T00:00:00+00:00</updated><id>https://marcodifrancesco.com/non-zurich</id><content type="html" xml:base="https://marcodifrancesco.com/2025-10-05-non-zurich/"><![CDATA[<p>Let’s explore the question: when applying to new positions, should I just apply to Zurich instead of all of Switzerland? It has been ~3 months since I started applying. This reflection comes because I got an interview from Basel a few days ago, and one of the first things I thought was “it’s way too close to the German border”. Here we go again with this mindset of diversification to slowly grind to the top.</p>

<p><strong>Past dream</strong>. Looking back at the past videos from Copenhagen, Sweden, and the Netherlands, the pattern is clear: I always prioritized Zurich compared to any other city in Europe. Even when in the Netherlands shit started going downhill, I filled the city ranking of cities I wanted to get a job in long-term, and the first was already Zurich, while Helsinki was in the non-considered section. I recall very clearly that my objective was to find a job in an international city, and Lugano I can definitely say is everything except for being international.</p>

<p><strong>Posting quantity</strong>. Let’s get some numbers: when I was in Copenhagen, 60% of the positions I applied for in Switzerland were in Zurich. The point is that there has always been a part of me which tries to maximize the chances of getting a new position and wants that missing 40%. I’m always thinking “well, if the company is good or the job is particularly good, I could stay there in that city for a while, right?”, and when asking my mom, she agrees with me. The problem with this approach is that by never narrowing down the amount of cities I’m applying to, I’m spending way more time applying to positions which are not in Zurich, and it’s time that I could potentially spend getting informed about companies in Zurich which I could join.</p>

<p><strong>Dream</strong>. Let’s picture the dream I have now. Picture if I’d move there, let’s say I am lucky enough to apply to a job, and next summer I’ll be there enjoying my summer weekends in the Zurichsee with Alberto and Federica. Will it feel good? Well, shit, thinking in this way it would be damn wonderful. Think of going out with their friends from the startup and meeting an Indian friend who is as motivated as you are in development. Think if I’d go out with them and I meet some girl whom I get along with. She doesn’t need to be my potential girlfriend, but just having a female girl with whom I’d go out from time to time with a group of friends would be wonderful. Think it even a level further, meeting some of the nerdy friends of Alberto and getting along with their group, drinking beer in the lake, playing volleyball in the fields, and just feeling like you are part of a group, and when you arrive they scream in Miran’s style “éeeee Marrrcoooo”. It feels like a kind of dream which you have been missing since Aalto, but it’s not that farfetched. I just need to find a good-ish position in Zurich regarding data science and I would feel like making some long-term investments in regards to relationships.</p>

<p><strong>Stability</strong>. Looking from that angle, going to Zurich compared to other cities would feel like a destination more than a stop. I’d feel like I can make a larger investment in terms of effort to find new friendships, a house, and activities. Of course I will feel like I want to change the job the first day I get there, but it’s inevitable, and I’d rather feel like I need to get a new job in the city I’m comfortable saying “I want to stay here for some years” compared to a city that, yet again, on day one, will feel like a stop.</p>

<p><strong>Now</strong>. You got to think long term. Go to Zurich, find a job which is fine, and feel for the first time like it’s your destination. Just damn learn all the possible companies in Zurich and make a small effort to contact 1 person of a company which is offering a job you are interested into, and ask him what he is working on. This is not to find a new job, but by listening at their experience you will get to know if you like and don’t like. Remember that you are in the period of your life where you have to dare to ask others for help. Probably most people will ignore you; at the end of the day, you are asking for the most expensive resource of this world: time. Eventually, you will get someone who shares their experience, and I’m definitely sure, after talking to them, I will picture me, Alberto, my sister, and some nerdy startup friends all together close to the beach volleyball fields, with Alberto, who explains some weirdass new paper he is writing, and I’ll feel part of something. During the Via Francigena this is what the Uruguayan girl said to do: do the impossible. This is impossible, and I’ll do it.</p>

<!-- 
## Other considerations

**Campus life**. In Aalto, everything was offered to me: amazing colleages, the guild, CLC, and the whole warm and welcoming environment. But that was campus life: a community designated and structured to feel welcoming. The univeristy was subsidized to provide this environment, new structures, free sports, etc. And it intuitively makes sense, at the univeristy young adults stay to at most five years and everyone is new. Every other large city like Lugano is built to function at scale, and if you want to build connections you have to earn them. You must go out to find a new sport, new firnds, and a new group for whatever you are interested into. You may think "well, the only difference is that it's not subsidized, but I can just pay", but that's not true. The city is already funn of people who are strictly interconnected and have enough people around them. You should seriously stop hoping to find another place like Aalto where everything was given to you, you were living campus life there.

**Language**: I don’t want to learn german and neither french. One thing I definitely know is that I don’t like to listen to french people speaking, while for german sometimes you think “yeah I could take back german”,  but I also know it’s fucking harsh and unpleasant to listen to, but I’d probably get used to it like you did with Danish, which is also damn harsh, and you seemingly didn’t get bothered by.

**Salary**. It would be very similar compared to other swiss cities in the german part, it’s not a real reason.
-->]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[Let’s explore the question: when applying to new positions, should I just apply to Zurich instead of all of Switzerland? It has been ~3 months since I started applying. This reflection comes because I got an interview from Basel a few days ago, and one of the first things I thought was “it’s way too close to the German border”. Here we go again with this mindset of diversification to slowly grind to the top.]]></summary></entry><entry><title type="html">Gold Step Back</title><link href="https://marcodifrancesco.com/2025-08-31-gold-step-back/" rel="alternate" type="text/html" title="Gold Step Back" /><published>2025-08-31T00:00:00+00:00</published><updated>2025-08-31T00:00:00+00:00</updated><id>https://marcodifrancesco.com/gold-step-back</id><content type="html" xml:base="https://marcodifrancesco.com/2025-08-31-gold-step-back/"><![CDATA[<p><strong>Gold step back</strong>: In may I got allured by gold returns in period 2000-2025 which have shown very similar returns and seemingly opposite trends. The thought of hedging against risk-adjusted returns was so fascinating that I allocated 12% of my 2025 end-of-year portfolio allocation.</p>

<p>The idea came when back in May I read the analysis by [earlyretirementnow] about gold allocation which he advised at 10-15%. Despite this advice he noted that he didn’t see enough pros in gold and remained 100% in equities. Like a real pro, I decided that after seeing the reddit post of the gif that compared the S&amp;P and GOLD trend in the last 20 years I went all in. No one can change my mind. Gold is definetly the way to go.</p>

<p><a href="/assets/img/2025-08-31-gold-step-back/reddit-sp500-vs-gold-1990-2023.mp4">reddit-sp500-vs-gold-1990-2023.mp4</a></p>

<p>And here I am, four months later, researching evidence to back my gold allocation. This time instead of looking in the last decade I wanted a larger horizon. Of course, what I found was overwhelming evevidence of exactly the opposite of that. This is perfectly summarized by [WisdomTree]:</p>

<blockquote>
  <p>Gold and stocks serve fundamentally different roles in a portfolio, and history makes their contrast crystal clear.</p>

  <p>Over the past two centuries, gold has been a relatively stable store of purchasing power, offering real returns of just 0.8% per year—barely ahead of inflation. It’s inert by design: <em>it yields nothing, it doesn’t grow, and it doesn’t compound</em>. Its primary function is wealth preservation, especially in times of political upheaval, monetary disorder or inflation shocks. In contrast, stocks represent dynamic claims on real assets—productive enterprises with reinvestment, growth and innovation embedded in their structure. From 1802 to April 2025, a diversified equity portfolio delivered real returns of nearly 7% annually, compounding purchasing power exponentially over generations.</p>

  <p><img src="/assets/img/2025-08-31-gold-step-back/gold-vs-stocks-1800-2021-inflation-adjusted.webp" alt="gold-vs-stocks-1800-2021-inflation-adjusted.webp" /></p>

</blockquote>

<p>But sure, let’s look at the opposite side. There is the paper [The role of gold in investment portfolios, Wagner, 2024] that like in the [earlyretirementnow] advises 17% gold, but they are doing it on a 60/40 equity/bonds portfolio where their target audience seems to be the average american which can pull out all their investment from a moment to another… I am 100% equity… What was I thinking.</p>

<p><strong>Stagflation</strong>: Btw, given the low profile audience the terminology was super easy and I finally get to learn what stagflation is, after hearing it many times in Ben Felix vidoes. It’s basically the situation where GDP falls and prices raise.</p>

<p><img src="/assets/img/2025-08-31-gold-step-back/stagflation-chart.webp" alt="stagflation-chart.webp" /></p>

<p>This word even though sounds super bad, it’s not as bad as deflation. Deflation is bad. Deflation happens very rarely (2% of the time) compared to stagflation (12%), and Wagner writes:</p>

<blockquote>
  <p>Deflation is a destructive situation in which economic activity and prices are falling. Japan serves as a  modern case study of the deleterious effects of this environment.  Governments generally attempt to avoid this situation at all costs, even  if that means printing money and risking an overshoot into higher-than-desired inflation levels.</p>

</blockquote>

<p>Interesting. At least I feel stupid for buying gold and intelligent for knowing a new word I can show off to my coworkers.</p>

<p>This is the new allocation, less go with stockss. I’m sure americanz know what they are doing and the market will grow to the staz and beyond.</p>

<p><img src="/assets/img/2025-08-31-gold-step-back/gold-new-allocation.webp" alt="gold-new-allocation.webp" /></p>]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[Gold step back: In may I got allured by gold returns in period 2000-2025 which have shown very similar returns and seemingly opposite trends. The thought of hedging against risk-adjusted returns was so fascinating that I allocated 12% of my 2025 end-of-year portfolio allocation.]]></summary></entry><entry><title type="html">Underwhelming Keyboard</title><link href="https://marcodifrancesco.com/2025-08-31-underwhelming-keyboard/" rel="alternate" type="text/html" title="Underwhelming Keyboard" /><published>2025-08-31T00:00:00+00:00</published><updated>2025-08-31T00:00:00+00:00</updated><id>https://marcodifrancesco.com/underwhelming-keyboard</id><content type="html" xml:base="https://marcodifrancesco.com/2025-08-31-underwhelming-keyboard/"><![CDATA[<p>Following the previous post 8 days ago: <a href="/2025-08-23-new-beautiful-keyboard">2025-08-23-new-beautiful-keyboard</a></p>

<p>Tried out LoFree Flow for 3 days now and from the first moment I received it I was underwhelmed. My priorities have definitely changed:</p>

<ul>
  <li><strong>Sound</strong>: it seriously sounds super wood-chopping-like in Stardew Valley. The switches are too silent, Tatiana even said it’s more silent than the office keyboard. I figured out I have a soft touch and can handle switches that can make more noise. The Nuphy Air v3 seems to be louder and a better choice.</li>
  <li><del>Blueooth 2-5 sec sleep recovery time:</del> Bullshit, the LoFree flow reconnects immediately. I don’t even have the time to click one key that it has already reconnected.</li>
  <li><strong>Programmable</strong>: Huge issue with LoFree Flow at the office is that I need to have the “Meta + hjkl” keys for the arrows. Moving right hand to the arrows in the office keyboard was intuitive since there was a gap between the main part of the keyboard (the rectangle from Ctrl to Del), and the four arrow keys. On the other hand with the contiguous keys in the 75% keyboard smash together the arrow keys with the ctrl and shift keys and make it much more difficult to have the muscle memory to click them.</li>
  <li><strong>Black vs white</strong>: the first keyboard was black I had with RGB backlights you can’t really see the backlighting easily, I’d rather try out the new white + shine through keycaps. Plus, I started this whole rabbit whole from “I want a beautiful white keyboard” and I should stick to it, not defaulting back to black as always.</li>
  <li><strong>Shine through keycaps</strong>: they are a must. Seriosly, today they are a must. Yesterday I didn’t care about them but today I can’t live without them. This morning (it’s 8:35, shit it’s early) I fell in love after seeing that the LoFree Flow v2 in white with white backlights, and shit if it looks good. Since today is the last day I give myself to buy the keyboard I will just take them. Fuck it, I got money, I need to spend them too.</li>
</ul>

<p>Conclusions:</p>

<ul>
  <li>I bought from the official website the Nuphy Air75 v3 with the white shine through keycaps (which yesterday were not available, it was destiny). I hope I don’t regret it since I spent 175$ after discount on it. Shit if it’s a lot of money.</li>
  <li>Only thing I’m afraid is that, like the LoFree Flow, it will be underwhelming when I’ll receive it. In the picture and youtube videos it looks so damn good, then you receive it and looks like an average keyboard with a metal frame. Let’s hope I won’t have that feeling.</li>
</ul>]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[Following the previous post 8 days ago: 2025-08-23-new-beautiful-keyboard]]></summary></entry><entry><title type="html">New Beautiful Keyboard</title><link href="https://marcodifrancesco.com/2025-08-23-new-beautiful-keyboard/" rel="alternate" type="text/html" title="New Beautiful Keyboard" /><published>2025-08-23T00:00:00+00:00</published><updated>2025-08-23T00:00:00+00:00</updated><id>https://marcodifrancesco.com/new-beautiful-keyboard</id><content type="html" xml:base="https://marcodifrancesco.com/2025-08-23-new-beautiful-keyboard/"><![CDATA[<p>In Aug 2025, after 4 months in Corner I have a sudden will to have a beautiful keyboard in the office. I want it so that if someone approaches me they say: “wow, that’s really a beautiful keyboard”.</p>

<p><strong>Requirements</strong>:</p>

<ul>
  <li>Beautiful.</li>
  <li>Low profile</li>
  <li>Silent: e.g. linear switches</li>
  <li>75%, 84 keys</li>
  <li>Color: fully white (if possible)</li>
  <li>Price: max 150 euros</li>
  <li>2.4 Ghz usb dongle to avoid 2-5 sec sleep recovery time (mentioned below)</li>
</ul>

<p><strong>Options by company</strong>:</p>

<ul>
  <li><strong>Lofree</strong>: most well known brand in youtube, they make by far the most beautiful keyboards
    <ul>
      <li>
        <p>Flow: most beautiful of them all: alluminium, beautiful edges, but unadvised to buy really an older model given the market is improving very quickly for low-profile keyboards.</p>

        <p><img src="/assets/img/2025-08-23-new-beautiful-keyboard/lofree-flow.webp" alt="image.png" width="250" /></p>
      </li>
      <li>
        <p><del>Flow v2</del>: horrible 2cm right edge, was better the v1 design. Relased in kickstarter in Aug 2025, and despite being super new, the right edge has everyone asking “why, just, why?!”</p>

        <p><img src="/assets/img/2025-08-23-new-beautiful-keyboard/lofree-flow-v2.webp" alt="image.png" width="250" /></p>
      </li>
      <li>
        <p>Flow lite: plastic which makes it 10% lighter, 20 euros cheaper, released in Sep 2024. Plastic makes it much less beatiful, for 20 euros less I don’t feel poor enough to buy it.</p>

        <p><img src="/assets/img/2025-08-23-new-beautiful-keyboard/lofree-flow-lite.webp" alt="image.png" width="300" /></p>
      </li>
    </ul>
  </li>
  <li><strong>Nuphy</strong>: the most recognized brand with Lofree.
    <ul>
      <li>
        <p>Air75 v3: loved among the reddit community which hates Lofree for lack of transparency (whatever it means). But seriously, I don’t need to handle to zoom on the top right of the keyboard.</p>

        <p><img src="/assets/img/2025-08-23-new-beautiful-keyboard/nuphy-air75-v3.webp" alt="image.png" width="250" /></p>
      </li>
    </ul>
  </li>
  <li><strong>Keykron</strong>: company seen as competition for the first two, at half of the price. Around $80 instead of the range $120-$160.
    <ul>
      <li>
        <p>K3: similar to what Bozzo had in FBK and looking much less like “professionally mac-looking beautiful white” and much more “I’m a computer scientist that types in the 10 keys method”.</p>

        <p><img src="/assets/img/2025-08-23-new-beautiful-keyboard/keykron.webp" alt="image.png" width="300" /></p>
      </li>
    </ul>
  </li>
</ul>

<p><strong>Lofree Flow overview:</strong></p>

<ul>
  <li>Switches overview:
    <ul>
      <li>Flow (white version): Ghost linear switches</li>
      <li>Flow lite (every color version): Specter switches, very similar to the Ghost switches</li>
      <li>Purchase separately: Hades switches, much more silent than the Specter/Ghost, compatible both with Flow and Flow lite.</li>
    </ul>
  </li>
  <li>If Ghost Linear Switches switches are too loud in the office I can buy for 60-70 euros the Hades switches. I highly doubt the noise is a problem given the noise Gabriele makes when typing and during the calls, and people from the other team scream a lot without anyone even complaining. But still good to have a silent keyboards for future offices.</li>
  <li>Bluetooth reconnection issue: after 10 minutes of inactivity the keyboard enters sleep mode and reconnection takes 2-5 seconds (some users say 1s, some 8s). This could be fixed by the 2.4 Ghz dongle which the Flow lacks, and only solution is to use the cabled connection at all time. Pro is that when it’s reconnecting the keys are remembered.
    <ul>
      <li>Bluetooth mode my perspective: the keys are remembered when it’s reconnecting. For example, if it goes to sleep after I loked the screen, I’ll be typing the password in ~3 seconds, by the time I hit enter the keyboard has waken up and in than instance it’s definetly usable.</li>
      <li>Cabled mode my perspective: even if it looks less shiny, keeping the cable plugged at all time may seem a reasonable tradeoff for its beauty. If I’m changing desk I’m anyway not taking the keyboard with me and that wouldn’t be an issue.</li>
    </ul>
  </li>
</ul>

<p><strong>Conclusions</strong>:</p>

<ul>
  <li>Lofree Flow is a no brainer after I found it for 144 euros on amazon fr that ships in Switzerland. Even considering the bluetooth connection issues, the first-sight love is still standing strong. Every partner has it’s flaws.</li>
</ul>]]></content><author><name>Marco Di Francesco</name></author><category term="post" /><summary type="html"><![CDATA[In Aug 2025, after 4 months in Corner I have a sudden will to have a beautiful keyboard in the office. I want it so that if someone approaches me they say: “wow, that’s really a beautiful keyboard”.]]></summary></entry><entry><title type="html">LightRiver — Master Thesis</title><link href="https://marcodifrancesco.com/LightRiver/" rel="alternate" type="text/html" title="LightRiver — Master Thesis" /><published>2024-08-01T00:00:00+00:00</published><updated>2024-08-01T00:00:00+00:00</updated><id>https://marcodifrancesco.com/light-river</id><content type="html" xml:base="https://marcodifrancesco.com/LightRiver/"><![CDATA[<ul>
  <li>Ported an Online Machine Learning model from Python to Rust, achieving a 97% reduction in execution time and attaining a processing capability of 10,000 records per second for IoT devices.</li>
  <li>Optimized memory usage to a minimal 20MB, facilitating deployment on robots with stringent memory limits and resulting in a 55% enhancement in positioning accuracy.</li>
  <li>Specialized implementation of Mondrian Forests, streamlining the model for real-time learning in performance-sensitive applications.</li>
</ul>

<h3 id="thesis--paper">Thesis — Paper</h3>

<p><a class="post-link" href="https://aaltodoc.aalto.fi/items/941f3205-3883-42b6-92f3-0d39840b654e" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 📜 Aalto Publication
</a>
<a class="" href=" https://purl.utwente.nl/essays/102624" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 📜 Twente Publication (mirror)
</a>
<br />
<a class="" href="/assets/img/light-river/master-thesis-paper.pdf" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 📜 Download PDF (mirror)
</a></p>

<h3 id="presentation">Presentation</h3>

<p><a class="post-link" href="/assets/img/light-river/master-thesis-presentation.pdf" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 🎤 Slides
</a></p>

<h3 id="github">GitHub</h3>

<p><a class="post-link" href="https://github.com/MarcoDiFrancesco/light-river-cache" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 🧑‍💻️ LightRiver-Cache
</a></p>

<p><a class="post-link" href="https://github.com/online-ml/light-river" target="_blank">
    <img src="/assets/redirect-icon.png" width="15px" style="filter: contrast(0);" /> 🧑‍💻️ LightRiver
</a></p>]]></content><author><name>Marco Di Francesco</name></author><category term="Machine-Learning" /><category term="Rust" /><summary type="html"><![CDATA[Ported an Online Machine Learning model from Python to Rust, achieving a 97% reduction in execution time and attaining a processing capability of 10,000 records per second for IoT devices. Optimized memory usage to a minimal 20MB, facilitating deployment on robots with stringent memory limits and resulting in a 55% enhancement in positioning accuracy. Specialized implementation of Mondrian Forests, streamlining the model for real-time learning in performance-sensitive applications.]]></summary></entry><entry><title type="html">NYCTaxis</title><link href="https://marcodifrancesco.com/NYCTaxis/" rel="alternate" type="text/html" title="NYCTaxis" /><published>2024-02-01T00:00:00+00:00</published><updated>2024-02-01T00:00:00+00:00</updated><id>https://marcodifrancesco.com/nyctaxis</id><content type="html" xml:base="https://marcodifrancesco.com/NYCTaxis/"><![CDATA[]]></content><author><name>Marco Di Francesco</name></author><category term="Data-Engineering" /><category term="Spark" /><summary type="html"><![CDATA[Analyzed taxi tipping behavior using Apache Spark to process large-scale datasets (67 GB of Yellow Taxi trip) to assess the impact of ride area, time, distance, and traffic on tipping. Employed Spark for data aggregation and analysis, generating insights into spatial and temporal patterns of taxi rides and their correlation with tipping trends. Visualized findings using Python libraries (Matplotlib and Seaborn), creating heatmaps and time-series graphs to depict variations in tipping across different NYC zones and times of the day.]]></summary></entry><entry><title type="html">MT-SMAC</title><link href="https://marcodifrancesco.com/MT-SMAC/" rel="alternate" type="text/html" title="MT-SMAC" /><published>2023-10-01T00:00:00+00:00</published><updated>2023-10-01T00:00:00+00:00</updated><id>https://marcodifrancesco.com/mt-smac</id><content type="html" xml:base="https://marcodifrancesco.com/MT-SMAC/"><![CDATA[]]></content><author><name>Marco Di Francesco</name></author><category term="Machine-Learning" /><category term="Hyperparameter-Optimization" /><summary type="html"><![CDATA[Implemented improvements to a multi-target Hyperparameter Optimization model, extending and enhancing the work of a PhD project. The project, known as MT-SMAC, focuses on the Sequential Model-based Algorithm Configuration (SMAC) for optimizing highly parameterized algorithms. Conducted empirical analysis using the YAHPO gym for benchmarking, providing insights into the efficiency and robustness of the model.]]></summary></entry><entry><title type="html">CourseMatch</title><link href="https://marcodifrancesco.com/CourseMatch/" rel="alternate" type="text/html" title="CourseMatch" /><published>2023-02-01T00:00:00+00:00</published><updated>2023-02-01T00:00:00+00:00</updated><id>https://marcodifrancesco.com/coursematch</id><content type="html" xml:base="https://marcodifrancesco.com/CourseMatch/"><![CDATA[]]></content><author><name>Marco Di Francesco</name></author><category term="Machine-Learning" /><category term="NLP" /><summary type="html"><![CDATA[Developed a Course Recommender System in a team of 3 students using a Transformer based Large Language Model (LLM). Utilized the BERT model from the HuggingFace library to generate embeddings for both the query and the corpus. Conducted comprehensive experiments on a custom-built database and evaluated the models using various similarity techniques. Recognized as the Best Project in the 2023 Natural Language Processing Course, surpassing 26 other projects.]]></summary></entry></feed>