My First Half Marathon🏃🏻

Sunday, July 26

드디어 잘 마쳤다! 아쉬웠던 건 언덕에서 생각보다 너무 퍼져서, 내리막에서 만회하고자 찍어누르듯이 달리다가 전에부터 신경쓰이던 발목이 또 도져서 끝까지 밀어붙이지는 못했다. 정말 정말 풀코스 뛰는 분들이 얼마나 대단한건지, 최근에 깨진 풀마라톤 Sub2기록이 얼마나 미친건지 새삼 깨닫게 되었다. 그래도 날씨가 정말 눈부시게 아름다웠고, 많은 사람들과 함께 뛰면서 받는 그 엄청난 에너지는 평생 잊지못할 것 같다. 사실 더 이상은 Half 이상의 대회를 나갈 마음은 싹 사라졌지만 ㅋㅋ 정말 후회없는, 보람차고 즐거운 경험이었다.

I finally finished! What was disappointing was that the hills wore me out much more than I expected. I tried to make up for it on the downhills, almost pounding my feet into the ground, but the ankle that had been bothering me flared up again, so I couldn’t push all the way to the end. It really hit me just how incredible people who run full marathons are, and how insane the recently broken sub-two-hour “marathon” barrier is. Still, the weather was dazzlingly beautiful, and I don’t think I’ll ever forget the tremendous energy of running alongside so many people. Honestly, any desire to enter another half marathon or longer race has completely disappeared lol, but it was a truly rewarding, enjoyable experience with no regrets.

Monday, July 27

CoreHPC로 이전을 성공적으로 했고, 본격적으로 Agent를 조금 더 활용하고자 적합한 harness를 생각하고 열심히 공부중이다!! 진짜 non-deterministic한 특성이 매력적이면서도, 참 어찌보면 예술에 가깝다는 생각을 하게된다.

I successfully moved to CoreHPC, and now I’m studying hard and thinking about the right harness so I can start making more use of agents!! Their non-deterministic nature is really fascinating, but it also makes me think that, in a way, this is almost an art.

Tuesday, July 28

ECHO라는 liquid handler 교육을 받았다! 이건 D300e랑은 다른 mech (Acoustic Droplet Ejection (ADE): Uses ultrasonic sound energy focused at the fluid meniscus to eject precise droplets from a source well to a destination plate suspended above) 으로 액체를 transfer하는데, 여러 well에 있는 liquid to liquid transfer가 되기떄문에 내 pipeline에 있으면 큰도움이 되기때문에 신청했었다. 앞으로 이 기기랑 빨리 친해지길.

I had training on a liquid handler called ECHO! It transfers liquid using a different mechanism from the D300e (Acoustic Droplet Ejection (ADE): Uses ultrasonic sound energy focused at the fluid meniscus to eject precise droplets from a source well to a destination plate suspended above). I signed up because it can transfer liquids from multiple wells, which would be a huge help in my pipeline. I hope I can get comfortable with this machine soon.

Wednesday, July 29

링크드인에서 본 감명 깊은 AI 관련 포스트가 있어서 공유한다. 박사과정에서 인공지능 도구의 사용 여부보다 이해와 독립적인 판단을 강조한다고 할 수 있겠는데 난 정말 공감한다. 갈수록 인공지능이 좋아지는건 자명한 사실이니 사용을 막는건 밑빠진독에 물붓는 상황이라고 생각한다. 단지 어떻게 활용해야 독립적인 연구자가 되는것이 목표인 phD에서 그 의미를 관철하며 잘 졸업할 수 있을지 치열하게 고민해야하지 않을까.

I’m sharing an AI-related post I saw on LinkedIn that really resonated with me. I’d say its point is that a PhD should emphasize understanding and independent judgment over whether someone uses AI tools, and I completely agree. It’s obvious that AI will keep getting better, so I think trying to stop people from using it is like pouring water into a bottomless bucket. Shouldn’t we instead think long and hard about how to use it while staying true to the purpose of a PhD—becoming an independent researcher—and successfully completing it?

Thursday, July 30

본격적으로 ECHO를 clonining에 사용하기위해 여럿 시도를 했다. 논문에 제대로 조건이 나와있지 않아서 여러개를 시도해보는데 생각보다 시간을 많이 쏟아야할 것 같다.. 흐 그래도 엄청 강력한 방법이니 꼭 꾸준히 시도해서 언능 내걸로 만드려고한다.

I made several attempts to start using ECHO for cloning. The paper doesn’t spell out the conditions clearly, so I’m trying different ones, but I think this is going to take more time than I expected… Sigh. Still, it’s such a powerful method that I want to keep at it and make it part of my own workflow as soon as possible.

Friday, July 31

요즘은 agent를 신나게 많이 쓰다보니 느끼는건 아 정말 산으로갈때도 있고, 모든 내용을 track하는게 정말 어렵다고 느낀다. 찾아보니 나만 이런 생각을하는게 아니라서, 여러 사람들이 다양한 방식으로 제어하는 skill들을 개발했다. 의도적으로 slow down을 시키는 느낌. 요부분을 열심히 공부하고 있는데 정말 재밌다. 진짜 내가 여러 사람이 되어서 동시에 일을 할 수 있는 느낌. 발전이 무섭기도 하면서 잘 올라타야한다는 생각이다. 프로젝트별로도 뭔가 딱 필요한 최상위 구조가 있고, 그 구조 안에서만 동작할 수 있게 한다면 좋을 것 같다.

I’ve been having a lot of fun using agents lately, but I’ve also realized that things can really go off the rails sometimes, and it’s so hard to keep track of everything. After looking into it, I found that I’m not the only one who feels this way, and people have developed skills to keep agents under control in all sorts of ways. It feels like deliberately slowing them down. I’ve been studying this a lot, and it’s really fun. It honestly feels like I’ve become several people who can work at once. The pace of progress is a little scary, but I also feel that I need to get on board. I think it would be good to have a well-defined top-level structure for each project and let the agents operate only within it.

Saturday, August 1

드디어 머리를 자르는날. 저번부터 새로운 바버샵에 다니는데, 오늘은 좀 과감하게 짦게 해보고 싶어서 부탁드렸는데… 아니 거의 두피가 보일정도로 skin fade를 해주셨다 ㅋㅋㅋ 단연코 군대에서 했던 머리보다 짧다. 근데 생각보다 편해서 나름 좋은것같기도 ㅎㅎ? 오후에는 virtual date로 Marigold cafe에서 먹었던 리코타치즈 + 꿀 + 사워도우 오픈샌드위치를 재현하고자 각자 오전부터 베이커리에 갔다. 난 아주 애정하는 타르틴 베이커리에서 어마어마하게 큰 사워도우 두개를 사고 (poppy seeds 파운드케이크도) 같이 맛있게 해먹었다!! 좀 쉬다가 핑계고 풍향중을 들으면서 집안일을 싹 하고 하루를 잘 마무리했다.

Finally, haircut day. I’ve been going to a new barbershop since last time, and today I asked for something a little shorter… Well, he gave me a skin fade so short that you can practically see my scalp😅. It’s definitely shorter than the haircut I had in the military. But it’s surprisingly comfortable, so maybe I actually like it haha? For a virtual date with my girlfriend in the afternoon, we each went to a bakery in the morning to recreate the ricotta cheese, honey, and sourdough open-faced sandwich we’d had at Marigold cafe. I went to Tartine Manufactory, a bakery I absolutely love, and bought two enormous sourdough loaves (and some poppy seed pound cake). We made our sandwiches together and really enjoyed them!! After resting a bit, I got all my housework done while listening to 핑계고 풍향중 and wrapped up the day nicely.