Skip to content ↓
avatar of MG: person with short blond hair wearing a blazer

a day in my life as a MISTI student by MG '24

ft. research and friends

As I’ve mentioned previously, I’m doing MISTI-Germany this summer and living in Berlin. MISTI is a really cool program that gives MIT students support to live and work abroad for a summer or a semester, without worrying too much about certain logistical aspects. I’ve always known that I would want to do MISTI; in fact, I was set to do it last summer, but things didn’t work out because of covid. I chose Germany because of my long-running interest in the country and its culture, as well as because, of all the languages I speak, German is the one that needs the most practice. I also wanted to spend a summer in Berlin specifically – it’s a very unique city with a character that appeals to me immensely. So, this summer I’m doing research at the meteorology institute of the Freie Universität Berlin and exploring the city in my free time. This blog is meant to be a slice of my life: I kept track of everything I did last Friday, and present it here for your enjoyment.

6:30 AM

I wake up at an ungodly early hour. Usually, I’ll get up around 8:30 – 9 AM, but today, I have some urgent tasks I need to finish before going in to work. Every Friday, my research group has a big meeting to discuss our progress, often with one or two presentations by individual members of the group to share what they’ve been working on. This Friday, it’s my turn to present the progress I’ve made in the month01 how has the time passed?? since I’ve arrived at the institute. It’s entirely on me for procrastinating, of course, but I had left a chunk of the presentation slides to be finished in the morning,02 which would have been more efficient than doing them late the previous night, so there’s that so I drag myself out of bed by 7 and get to work.

8:00 AM

By 8, I’m done with the presentation, so I let myself sleep for another half hour. I get up at 8:30, pack my “lunch”,03 three apricots and the leftovers of one of those meal replacement shakes from the previous day… yes I’m doing fine why do you ask and am out the door before 9 AM.

9:00 AM

My commute to the institute is honestly a little gross. First, I need to catch a bus, which stops across the street from my apartment. Sometimes, I get lucky and I only wait for a couple minutes; other days, it can be ten to fifteen minutes of standing at the stop. Generally, though, the bus runs really well, and the vehicles themselves are nice and new. I take the bus for about twenty minutes, then get off where the route intersects with a suburban-bound train line.04 the S-Bahn, these trains run above ground, in contrast with the U-Bahn, which runs underground I never wait more than 8 minutes for the train to arrive, and the trip itself is around ten minutes. I get off in Steglitz, a relatively quiet and residential area on the outskirts of Berlin, home to the Freie Universität Berlin. The university buildings themselves are pretty dispersed, and I’ve never seen the main campus.05 I suppose I probably should before I leave It’s a ten-minute walk from the train station directly to the meteorology institute, and I go directly there.

pastry, coffee, and apricots at a computer desk

my breakfast…

9:45 AM

On the way to the institute, I pick up a pastry for breakfast. I do this most mornings when I commute, so I cycle through the different pastries I like. Today I’m feeling a “quarktasche”, which consists of salty flaky dough filled with sweet cottage cheese. I’ve also brought a coffee drink thing from home.06 I stock up at the supermarket, it’s cheaper than buying a latte every day, though not cheaper than not having a caffeine addiction altogether With my little breakfast in hand, I take the fifteen minutes I have before my meeting to go over my slides and prepare myself to present.

10:00 AM

At 10, I head over to the meeting room. I’m presenting second, which I don’t mind. The first presentation is by an undergrad meteorology student, whose senior thesis is focused on analyzing data from a new crowd-sourcing initiative by the DWD.07 the German weather service The presentation is in German, so I only catch bits and pieces of it, but it seems interesting. Generally, the group I’m working in has a number of international grad students and postdocs, so most of the meetings and presentations are conducted in English. This particular one is an exception, since the presenter is an undergrad student, and the undergrad program is taught entirely in German.

After the first presentation is over, it’s my turn to present. I’m a little nervous, since I haven’t had time to run through it fully. Usually, I prepare much more for presentations, it’s just that this week my work got away from me. Plus, the presentation is meant to be an informal progress report: intended more so to solicit advice than to deliver results.08 because I don’t have results oops This summer, I’m working on a project to forecast Berlin fire brigade operations based on radar precipitation data. I’m specifically looking at “water damage”-related operations, which generally arise from flooding due to rain. Previously, some of the members of the group have developed a statistical model that, given radar data and a map of building density, can assign a probability of a fire brigade operation in any given area. My job this summer is to develop a machine learning model that can produce a similar forecast. The hope is that applying machine learning would reveal unexpected connections that a straightforward statistical method is unable of identifying.

laptop screen with slide titled "model doesn't work"

as you can see, my project is going swimmingly

The problem is, I’m beginning to suspect that the machine learning approach is overkill for the problem and datasets at hand. In the month that I’ve been wrestling with the problem, I haven’t been having a lot of luck with getting the model to produce a useful result. To make things even harder, the group I am working with is the “statistical meteorology working group,” meaning they focus on statistical methods exclusively. I was brought in as a sort of external machine learning “expert”, to see what I could do with their existing data. Now, I’m very far from an expert in machine learning. I’m not inexperienced, by any means, I’ve been doing ML-related research for a while now, and I took 6.03609 intro to machine learning, now called 6.3900 and did pretty well in it, but I’m at the level where if something goes wrong, my ability to troubleshoot is quite limited. And a lot of things have been going wrong. Consequently, a lot of my MISTI work lately has just been me doing a lot of googling, spending hours paging through Stack Exchange for ideas.

I mention all of this, in different words, during my presentation. I describe every idea I’ve tried, and the ways in which it has failed, and then present the approaches I’m in the middle of implementing. The presentation goes well, especially given how last minute I prepared it, and the discussion period that follows it is very productive. I am reminded of why I like science so much: I really enjoy bouncing ideas around, seeing everyone’s different approaches to problem solving, and trying to piece together a solution to the question at hand. I walk away from the meeting with a page-long list in my research notebook of things to try in an effort to get the model to work.

computer screens, hand holding a drink, and a book

my lunch. I thought this photo was really funny. I only got like two pages of reading done but that doesn’t matter

12:30 PM

After the meeting, I go back to my office10 I say office, it’s just a room with a bunch of computers that I’ve been using as an office since it’s not needed in the summer to decompress and have lunch. I typically alternate days working at the institute and working remotely from home.11 since the commute is so annoying On the days that I work from home, I like to cook to break up the monotony of work; I’ll usually make enough to take the leftovers with me to the office the following day. It’s a good cycle, but I broke it this week by going into work in an irregular pattern and also not having the energy to cook. So, for the second day in a row, I’m drinking one of those meal replacement shakes instead of lunch. I swear to myself that I’ll actually cook dinner today.

1:00 PM

After spending an unfortunate amount of time scrolling on Instagram, I get back to work. I take stock of the morning’s discussions and sort out my list of things to try for the model. It’s a good list, and it gets me excited for work on Monday, but I don’t really feel like starting any of the items now because none of them are very quick. I have a meeting set with my supervisor for 2 PM, so I poke around my data a little bit while I wait.

2:00 PM

The meeting with my supervisor goes well. We mostly go over the morning’s meeting, and he helps me make adjustments to the list of things to try. We also talk about my goals for the rest of the internship, and the logistical details of the end of the summer. Turns out, he’s going to be on vacation for the last two weeks of my internship, so I’ll probably end up giving my final presentation remotely at a later date.

2:45 PM

After leaving the meeting, I finish up the data-poking I had started after lunch. I decide that, even if I stay at work until 5 PM, I’m not going to be productive for the rest of the day, so I might as well head home. I’ve generally found, in all the internships that I’ve done, that research has a very relaxed approach to working hours. No one keeps close track of when I work, with the expectation that I’ll get all my tasks done and make progress, just at a time convenient to me. It’s true – all the scientists I know work odd hours, though this is also particularly a function of the sort of science12 computational or mathematical science that doesn’t require lab work or timed experiments I am immersed in. In any case, I personally enjoy this flexible approach to work, and find that it mostly works for me. It has its pros and cons, though, the con being that I never exactly feel like I’m not at work, since I could do my work at any time. I personally struggle with this feeling a fair amount, but it is one I think I can learn to live with and manage in a healthy manner.

view of city from train

more commuter views from my way home

3:00 PM

I consider stopping by somewhere on the way home, just for fun – a bookshop or a café or something, but decide against it. Amber is coming to visit for the weekend, and her plane lands at 5:30 PM, so I want to clean up the mess that is my apartment before she gets there. I also need to do my laundry, or I’ll have nothing to wear in the evening. So, I head directly home, reading13 I have the Murderbot Diaries series downloaded as ebooks, and it's perfect commuter reading on my phone while in the bus and train.

3:45 PM

When I get home, I vegetate for a while on my couch, scroll through Instagram, and respond to texts. Commuting is honestly so exhausting, and I can’t imagine having to do it every day. I’m glad I’m learning this now. After fifteen or so minutes, I decide to get up from the couch and embark on a cleaning spree. I pick up the things I have strewn around my room, get the laundry going, then go to take a shower.

5:00 PM

As I wait for Amber to get from the airport to my apartment, I start cooking dinner. I have a number of ingredients that I feel are on the border of going bad, so I just make something that incorporates them all: fried potatoes, sauteed mushrooms with yogurt, braised cabbage, and some eggs for added protein. I’m trying out my mom’s braised cabbage recipe for the first time, so I’m a little nervous that I’m going to mess it up, but it goes pretty well. For the mushrooms, I had found fresh chanterelle mushrooms being sold in my local grocery store, and immediately knew I had to get them. I have fond memories of my grandfather, back in Russia, bringing back baskets full of chanterelles from his mushroom-picking expeditions in the woods near our house. I’ve never seen them sold fresh in the US, though there’s a good chance I just don’t know where to look. As I cook, I put on the next episode of the new Star Trek show, which has lately been bringing me a lot of joy.

6:30 PM

After some delays with her flight and getting to my place, Amber finally arrives. I haven’t seen her since the end of the semester, so it’s a long-awaited reunion. I’m also really excited to show her Berlin and all the parts of it that I’ve been falling in love with. We eat dinner and catch up, then make plans for what to do for the rest of the night.

7:30 PM

I decide that we should go to the East Side Gallery, which is a kilometer-long section of the Berlin wall that has been preserved and covered in murals. I think, of all the tourist-y sites in Berlin, this one is particularly worth seeing, just because the murals are so cool and beautiful while also carrying a lot of meaning and history. Amber and I walk through the “gallery” and wander around a little around the area. We don’t stay for too long, since we have plans to meet Nanako K. ’25, who is also doing MISTI in Berlin, at a bar around 10 PM.

9:15 PM

We get back to my apartment and start getting ready for the night. We have plans to meet Nanako, go to a bar for a bit, then head over to a dance club for the rest of the night. Amber and I are both chronically late to things, and Nanako generally isn’t, so I make an effort to try to be speedy in my getting ready. We still leave too late to be on time, but that’s okay.

10:15 PM

The bar doesn’t quite pan out, so we decide to just grab something from a “späti” – a little corner store that carries convenience items like snacks and drinks – and sit outside. The shopkeeper asks me where we’re from, and I say we’re from the US. He tells me that he gets a lot of Americans in his shop, because, if I understand this correctly, the CIA has an academy nearby for new recruits to perfect their German. Why not. The weather is lovely, and Nanako, Amber, and I sit outside the späti and enjoy the night, before heading over to the club.

amber smiling at a subway station

Amber in Berlin!

11:00 PM

Berlin is known for its vibrant nightlife scene, and particularly for its techno clubs. I’ve never been particularly into techno or any other electronic music, but in Berlin, it makes a lot of sense. The clubs are great, the vibes are good, and dancing to techno with a big crowd is really fun. Tonight, we end up at a club I haven’t been to yet. It’s alright – it turns out to be a slightly older crowd than we’re used to, but there’s some young people mixed in, and the whole place carries a very unique character.

12:30 AM

We leave earlier than I typically would, but everyone has had long days, and the lack of sleep from the previous night is definitely hitting me. Amber and I part ways with Nanako, and I ask her to text me when she gets home. As we leave, I promise Amber that tomorrow we’ll go to a different club, one of my favorites, and it’ll be an even better experience. We’re back at my place and in bed by 1:30, ready to sleep in and go out into the city again tomorrow.

  1. how has the time passed?? back to text
  2. which would have been more efficient than doing them late the previous night, so there’s that back to text
  3. three apricots and the leftovers of one of those meal replacement shakes from the previous day… yes I’m doing fine why do you ask back to text
  4. the S-Bahn, these trains run above ground, in contrast with the U-Bahn, which runs underground back to text
  5. I suppose I probably should before I leave back to text
  6. I stock up at the supermarket, it’s cheaper than buying a latte every day, though not cheaper than not having a caffeine addiction altogether back to text
  7. the German weather service back to text
  8. because I don’t have results oops back to text
  9. intro to machine learning, now called 6.3900 back to text
  10. I say office, it’s just a room with a bunch of computers that I’ve been using as an office since it’s not needed in the summer back to text
  11. since the commute is so annoying back to text
  12. computational or mathematical science that doesn’t require lab work or timed experiments back to text
  13. I have the Murderbot Diaries series downloaded as ebooks, and it's perfect commuter reading back to text