Skip to content

My AI MBA Journey at the University of São Paulo: Triumphs, Struggles, and Everything in Between

The Search for Knowledge (and a Fatter Paycheck)I recently completed my MBA in Big Data and Artificial Intelligence at the University of São Paulo (USP). After finishing my undergraduate degree and taking some well-deserved time off from mathematical calculations and the insane routine of juggling college and work, I decided it was time to level up my career. Let’s face it - these days, just having a bachelor’s degree isn’t enough to climb the ladder (or fatten your wallet).My first challenge was deciding what to pursue. I had been flirting with AI for some time and was already working with Big Data. The Google algorithms, fed by all the digital breadcrumbs we leave across the internet, helpfully steered me in this direction. After searching for “MBA and technology,” AI programs popped up as sponsored results. I read through the curricula and decided this field made sense for my next step.The second step was finding the institution with the best program for my needs. I was focused on finding a course with more AI theory and practice than actual business management (yes, I’m aware MBA stands for Master of BUSINESS Administration - sometimes my brilliance knows no bounds). I researched options at PUC, Mackenzie, and USP, knowing that the university’s reputation would carry weight on my resume. Fortunately, my salary at the time allowed me to consider the cream of the crop. I ultimately chose USP’s online MBA program with a duration of a year and a half – because who doesn’t want to extend their suffering?## The Program Structure: Not Your Average MBAThe curriculum features 9-10 core modules plus 3-4 elective specialization modules in areas like audio applications, text, video, image processing, data visualization, and data center architecture - a significant differentiator. The basic modules range from Python and pandas fundamentals to creating those historical data videos everyone posts on LinkedIn (you know, the ones that somehow convince us that watching colored bars race across a screen for three minutes is a productive use of our time – and yes, I’m hopelessly addicted to them).The classes are pre-recorded, but each module includes live therapy/mentoring sessions with professors for those with questions. “Therapy” is the right word, because you’ll need emotional support after some of these modules. Another interesting requirement is an extension module that requires you to watch at least 7 online lectures or write summaries about them if you can’t attend live. These fascinating lectures about the AI universe are all available on YouTube for anyone to view. When lectures are live, they share a Google Forms link to mark attendance and secure credit. I appreciated how this approach pulled students away from pure theory and calculations, something I suffered through until my final undergraduate semester when we had legislation and workplace safety courses (though I nearly failed legislation because I had already transformed into a cold, calculating human being – engineering does that to you).With the program selected, I faced USP’s selection process. It’s not overly grueling - you submit your resume and work history for analysis to ensure you’re not wasting money on a course you can’t handle. Let’s be blunt: AI is essentially applied statistics, which is a mathematical field, combined with programming. With my background in electronic engineering and programming experience, I was accepted. And since I’m being completely transparent - and this information is publicly available on the website - in 2023, I paid around 23,000 Brazilian reais for this course. Yes, that’s right – I voluntarily traded a perfectly good used car for the privilege of staring at my computer screen for hours on end.## The Perks and Structure: What You’re Getting for Your MoneyThe program offers several perks: one year of full access to Alura (a learning platform), a USP Gmail domain (because nothing says “I’m educated” like an email address ending in .usp.br), a USP student ID (providing discounts and access to university infrastructure like gyms, libraries, and university restaurants), GitHub Pro with Copilot (so an AI can write code for you while you learn about AI – oh, the irony), and access to classes up to one year after completing the course (because you’ll definitely need to rewatch everything when you realize you retained about 5% of the content).Modules are divided into two-week segments, each with an assessment for content assimilation, and each module ends with a comprehensive exam. Both are multiple-choice with no time limit but have specified completion dates. The bi-weekly tests have 5 questions with unlimited attempts until you achieve a perfect score (for perfectionists or those who want to humblebrag on LinkedIn about mastering the content with their superhuman intellect). The end-of-module exams have 10 questions with only 2 attempts (though students have WhatsApp groups where they share answers faster than gossip spreads in a small town). As an adult who paid the price of a used economy car for this course, it’s your call whether to cheat or not. Your moral compass vs. your desperate need to pass – a true ethical dilemma of our times.### Module Structure- Bi-weekly segments

  • 5-question assessments (unlimited attempts)
  • 10-question module exams (2 attempts)
  • Average video length: 25 minutes
  • Some modules contain up to 50 videos (miss a few and you’re screwed)The module lectures are recorded and uploaded to Vimeo, but the entire system is proprietary to USP. The content is well-organized with downloadable materials (which shouldn’t be shared, for ethical, legal, and respectful reasons), submission areas for exercises, and question forums.Lecture time varies between professors, averaging around 25 minutes per video. That might seem brief, but some modules have up to 50 videos - fall behind and you’re screwed.The content quality is excellent. The professors are industry experts with incredible academic credentials (it’s USP, after all – they practically require you to have three PhDs and a Nobel Prize nomination to teach there) and delve deeply into theory. Even with my engineering degree from a good university (including Calculus 1-3, Physics 1-4, etc.), I struggled to keep up with the more in-depth material - but that’s exactly what I was looking for. Someone to make me feel intellectually inadequate again, just like in my undergraduate days.All theoretical content includes bibliography for further study (as if you’ll have time for that). Beyond theory, there’s practical application. Some professors alternate between theoretical and practical lessons, while others cover all theory before moving to practice – presumably to ensure you’re thoroughly confused before you try to write any code. The practical component involves developing/following algorithms in Python on Google Colab, applying the theory. I found this incredible and gained real hands-on AI experience. All notebooks are available for download too, so you can revisit your failures at your leisure.## Thesis Work: The Real TestHalfway through the program, the first thesis course begins, with four courses total. This first course teaches how to properly write a thesis - researching and referencing bibliography, structuring text, making citations, etc. This might seem basic, but not everyone knows how to properly cite sources (shocking, I know). They also teach LaTeX and provide USP templates – because apparently writing a thesis isn’t painful enough without also learning a markup language that looks like it was designed by a sadistic mathematician.The thesis process spans about 8 months. You choose your topic and submit it to the program coordinators, who assign you an advisor. I was incredibly fortunate to have Professor Jean Ponciano, who provided tremendous help throughout the process (a genuine saint among the academic demons). The thesis is submitted in written form, with incremental progress required for each project module. Ideally, by the end of these four modules, you’ll have completed your entire written project, making the process less traumatic (though in practice, theory and reality often diverge, much like my sleep schedule and my study plan). Each submission is reviewed, and corrections may be requested – “may” being the understatement of the century.Finally, the thesis is validated after a 10-minute presentation to a committee, open to anyone who wishes to attend. In my year, this was conducted via Zoom, mercifully sparing us from having to dress professionally below the waist.## My Personal Experience: The RollercoasterLike 99% of students, I started out doing everything perfectly - following the schedule, completing assignments on time, and dedicating about 2 hours daily, 7 days a week. I was basically the poster child for online education discipline. Around the third module, I began to lose focus, which screwed me over for the rest of the course. Everything snowballed; I had to skip some classes and tests, and things never got back on track – rather like watching one YouTube video and suddenly it’s 3 AM and you’re learning about conspiracy theories involving the Illuminati and garden gnomes.I thought I had enough maturity to be disciplined in an online course, but it’s incredibly difficult. It’s also exhausting. On days when my job was demanding, I had no mental capacity left to understand the mathematical algorithms of a deep neural network - and I don’t even have children! (Those of you with kids who completed this course deserve a medal, or therapy, or both.)I ended up focusing on modules that sparked my curiosity, continuing this way until the end of the course, which left me feeling frustrated with myself. I’m now retaking my MBA at a more relaxed pace, taking advantage of the one-year access to content post-graduation. Second time’s the charm, right? (Or at least that’s what I keep telling myself as I watch the same lecture for the third time.)### Reality Check- Started strong: 2 hours daily, 7 days a week
  • Derailed around module 3
  • Never fully recovered my study rhythm
  • Currently retaking courses during my 1-year access periodAt this point, you might think I didn’t learn a damn thing, but that’s where you’re wrong. I did the same during college, but I completed all exercise lists and assignments, which translated into excellent grades and helped me assimilate content better than just staring at professor’s slides like a zombie. That’s my learning style - I receive a problem I have no idea how to solve, panic, Google frantically, study how to solve it, and end up diving deep into the subject. Efficient? Debatable. Stressful? Absolutely.Since each module has one or more professors who change with each module, teaching styles vary as well. Not everything was perfect - 2 or 3 modules were extremely difficult to follow, one featuring 45+ minute lectures on data cubes and dimensional abstraction in databases (I gave up because I couldn’t understand a shit).## The Thesis: My RedemptionSince I hadn’t dedicated much time to the videos, I wanted to create an outstanding thesis that could potentially help me get into a master’s program. I’m quite perfectionistic about such things, which often works against me, but it’s the price I pay to try to achieve my goals (that, and the therapy bills).My thesis took about 8 months. I chose to apply a data drift detection algorithm (Page-Hinkley) to photovoltaic plant inverter data to identify failures and performance losses from factors the inverter couldn’t detect. Throughout this process, I faced numerous challenges in proving my algorithm was functional - not just technically, but also in finding subject matter experts to validate the project who weren’t complete jerks.I won’t delve too deeply into this part, but I had terrible experiences with “experts” who seemed more interested in criticism than assistance, which nearly crushed me. These people apparently got their PhDs in “Making Others Feel Inadequate.” Thanks to my advisor, Professor Jean Ponciano, and the emotional support he provided (seriously, the man deserves sainthood), I didn’t give up and ended up producing one of the best works in my class. This experience taught me to trust myself more and to be selective about whose opinion I value – a lesson arguably worth more than the MBA itself.Finally, I presented my project and received incredible feedback from the committee on the project, writing, and presentation, validating my goal of potentially using this work as a stepping stone to a Master’s program. The validation almost made up for all the sleepless nights and existential crises. Almost.## The Takeaway: Worth It, But Not EasyI learned a great deal about AI and feel genuinely prepared to implement solutions and delve into a Master’s program (though at a more measured pace, because I’m not a complete masochist). Obviously, compared to the professors, I’m still very much a junior among juniors – more like a fetus in the AI world.In summary, it wasn’t easy going through this journey, but it was a tremendous learning experience that I recommend to everyone interested in the field who has 23,000 reais they don’t need and a high tolerance for pain. I’ll describe my project in detail in another post - this one is already approaching TL;DR territory, much like most of my academic papers.