Day 62: 4 Weeks to Go!

The Final Countdown

It’s crazy to think that there are only 4 weeks left to go. By June 28th, I will have stayed in Germany for exactly 90 days. Time has simultaneously flown by and slowed down at the same time. I’m really glad I have this growing archive of posts to chronicle my adventures and experiences.

Routine

It was a rough start to the day. I didn’t fall asleep till late. I was too busy hacking out my lungs coughing. The latte macchiato from the bakery helped wake me up as I drove to the office. The weather has been particularly warm and beautiful these past few days, so I drove in with the windows.

I quickly regretted that as pollen invaded my nose. I was sneezing all the way till lunch. I forgot that there’s so many forest areas and fields all over. I’m definitely not used to that compared to the US. It was business as usual in the office.

Evening

I wrapped up at the office at around 7:15PM or so. When I got home, I felt like I had the energy to go for a run miraculously. But I knew better than to go for a run now. It was the euphoria that creeps in being awake for too long influencing my better judgment. I learned the hard way not to give in. The body is more tired than it seems and that’s what leads to injuries.

Instead, I opted to go for a walk to Five Guys. I wanted to stretch my legs and enjoy the weather, but I also craved something from home. I will say, Five Guys in Germany is pretty darn close to the real thing. I had a Bacon & Cheese Hot Dog. I made it German by adding mayo. It was delicious. I also had an Oreo Cookie Shake, which hit the spot 👌🏾. I was joking with my colleagues today that it’s only a matter of time before Chipotle makes its way to Germany. I think there’s a real market opportunity there.

LLMOps

Making my way back to the apartment, I jumped back into to exploring LLMs in more depth. In addition to my ongoing ventures into Stephen Wolfram’s article, I came across a great article on Weights & Biases on the new topic of LLMOps. The article describes the growing need to have specialized machine learning operations (MLOps) tools that cater to developing, scaling and deploying LLMs. I love the term.

From my activities at work, I’m coming to appreciate looking through the lens of infrastructure and maintenance when it comes to building tremendously complex systems, and going beyond the perspective of functional architectures. The complexity of scaling something is almost always hidden away in the backend. I say that from first-hand (and currently lived) experiences.

I also spent some time quickly glancing at a new course that’s being launched by Khan Academy called AI for Education. There’s only 1 unit so far, but the articles and experiments cited led me to the website of Ethan Mollick, a professor from the Wharton School at UPenn. He’s spent a lot of time investigating GenAI, and LLMs in particular.

I find his experiments figuring out how AI can be an insightful tool not just for learners but for educators to be really intriguing. His most recent article was really thought-provoking. I’ll spend some time playing around with Claude once I get access. It’s time to now hit the bed.

Frieden und Liebe ❤️ (translation: peace and love)