Minsing Log
Go back

English Edition

2024 Memoir

Founder Notes

There are less than 2 hours left before 2024 ends. I would like to conduct this year’s long-postponed year-end review.

🎯 Comparison with Early-Year Goals

Recalling my goals from the beginning of the year, they can be largely divided into three categories:

  1. Confidence that I can implement anything
  2. Understanding people’s needs through business experiments
  3. Financial independence from my parents

And my current, changed goals are:

  1. Confidence that I can implement anything
  2. Definition of work: Distinguishing between work I like and work I admire, and my strengths (grit)
  3. Training to grasp the core, no note-taking
  4. Becoming desensitized to fear
  5. The experience of finding a perfect problem and solving it
  6. The ability to think while executing once an idea pops up
  7. Finding my true calling. Trying out work to find what I am good at
  8. Not solving inefficiency with diligence. Always looking things up before doing them
  9. Overseas experience
  10. Inevitably using dependencies when doing projects
  11. Small-scale exit, sharp product
  12. Leading a project with initiative
  13. Trying and experiencing as much as possible.
  14. Breaking away from perfectionism
  15. Growth solutions, goals, and growth plans truly suited for the ‘real me’
  16. Growth in development
  17. Routines - reading, retrospective, English without subtitles
  18. Knowing myself
  19. Growing my basis and building expertise
  20. Turning what I like into a nomad lifestyle abroad, networking, earning foreign currency, but not chasing money
  21. Converting Bitcoin assets into foreign currency
  22. Two-month intensive retreat, building one thing every 2 weeks

Definition of work, definition of what I like

⌚️ Timeline of Major Events

1️⃣st Quarter

Before the semester started, I tried out various challenges here and there.

✨ RAGChain: Participating in my first-ever open-source project

After being discharged from the military, I came to Garage with the mindset of doing nothing but development in an intensive retreat, and blindly tagged along. Those three months were when my development skills improved the fastest this year. It was a time of constantly learning, applying new things, and receiving code reviews. It was a project aimed at overcoming the limitations of LangChain, but LlamaIndex execution speeds were very fast and already well-built, so we pivoted to AutoRAG. The initial concept was RAG evaluation, but through stacking innovations, it evolved into concepts of easy RAG optimization and fast RAG building.

What I could focus on best was the intensive retreat aspect—setting aside all other schedules and focusing solely on development and reading. What was regrettable was the lack of rest—I rested without really knowing how to rest properly—and that the older guys I was doing the project with were in school, so we couldn’t work together consistently.

💪 Before & After

💡 KPMG Ideathon: The first project I led myself

After AutoRAG, left behind when I departed for Portugal as a nomad, I wanted to try something new, so I gathered people around the theme of a hackathon. It was my first challenge this year going against my nature. As much as there were regrets, it was a challenge from which I learned a lot. I was genuinely terrified of leading work alongside people better than me. I learned that bumping into things and moving forward is much more important.

💪 Before & After

📞 AI Call Center Agent Observability

This was my first project with T-hyung and Dr. Cheol. Although we proceeded with the thought of starting a business, unfortunately, it didn’t lead to an actual startup.

💪 Before & After

🤼‍ First AI Developer Networking

I attended a total of two networking events. The reason I wanted to do this was to eliminate my fear of new people and improve my speaking skills. The very first networking event was held at Maru 180. It was a place where companies explained their solutions, we asked questions about generative AI, and we networked. LLMs were starting to become a major issue, and new concepts like prompt engineering and AI villages were emerging. Free pizza, free desserts, and beer tasted sweet. Second, I participated in a generative AI networking event held at Nonce. It was a paid networking event, but it was engaging and fun. It was a networking event where I dived in headfirst without knowing much, but standing and listening to stories from people in the same industry was fascinating.

💪 Before & After

📗 The Beginning of a Book Reading Club

Having experienced a leading role in the ideathon and feeling like I went against my nature and took on a challenge facing my fears, I wanted to take the next step. That was a book club with the purposes of:

  1. Playing a leading role based on what I am good at and enjoy
  2. Reading books, solving problems, and explaining them to make them truly my own
  3. Gaining insights through others reading for me, since I can’t read every book myself
  4. A comfortable club where we freely read and chat, centered around reading as a process of solving life’s problems.
  5. Having people who want to build a habit borrow my energy of consistently reading to read along together.

After talking with Yongjin, I became convinced it was worth a shot. Gradually gathering people from acquaintances, it now maintains around 5 to 6 members. I am thinking of expanding the scale in the new year and renewing the concept into a theme for building good habit routines.

💪 Before & After

2️⃣nd Quarter

From this point on, rather than development techniques, I focused on deepening text mining and visualization utilizing LLMs.

📰 Participating in the Korea Research Public Opinion Analyzer Project

We conducted in-depth news analysis utilizing LLMs. The main feature was building a tool that could analyze public opinion by processing a large volume of news text using LLMs and visualization tools. It was an experience of applying the OOP concepts I had tried in the previous RAGChain project all by myself. In the later stages, since there was no special evaluation metric for whether the analysis went well or not, struggling through human evaluation of the analysis quality one by one using 100 sample news articles was tough, but quite meaningful.

It was fun to brainstorm and define meaningful analysis with T-hyung and Dr. Cheol while researching, experimenting with, and analyzing news articles and comments separately. Previously, I thought LLMs were almighty AIs that could handle anything, but upon using them, I found areas that needed improvement and aspects that were difficult to replace humans with. First, since answers lacked consistency, exception handling and prompt engineering were needed to ensure consistency in analysis and answers. Second, it was still difficult to trust the content judged and analyzed by LLMs during analysis, and there was the difficulty of lacking an appropriate evaluation metric to determine whether an analysis was good or bad. Third, creating train sets, test sets, and correct-answer datasets to judge answer quality required making various datasets for analysis, which became a bottleneck in terms of time, leading us to push it to the next step. Consequently, at this point, I began to see LLMs as auxiliary tools to reduce human simple labor.

💪 Before & After

📀 Meeting T-hyung and Almost Developing a SORA AI Video Generation Service

My first experience participating in a pre-startup project. As Dr. Cheol advised, he recommended a hyung who seemed like a good match, and we built chemistry while working on the call center project together. Communication issues were quite significant, but we tried to align as we progressed further with the project. Jgj-hyung, a frontend developer, joined, and we ideated items in preparation for the upcoming release of SORA. The idea started as AI meme generation, and later evolved into version control for AI video prompts and outputs in order to produce AI videos.

However, as meetings continued, repeated, and dragged on, communication lost its focus, and team members continuously expressed dissatisfaction, causing me to lose energy and eventually leave the team while attending school. T-hyung’s situation was that he was busy with corporate projects just like me, and needed funds to run the runway. Jgj-hyung was facing the job-hunting season while lacking funds to run the runway. In my situation, it was difficult to focus on a startup while juggling school and the public opinion analyzer corporate project. It was extremely difficult for all three of us to invest the same energy and time, and gather at promised times. Meetings were also fruitless. Meetings always went off track, and the reasons for going off track were mainly focused on talking about C and D instead of agenda A, venting hurt feelings, or expressing dissatisfaction with meetings rather than discussing problems.

What I felt was important to proceed with a startup in this project were:

  1. Sticking together at fixed and promised times for necessary situations and pouring in time and energy
  2. Fast execution, rapid validation of ideas
  3. Reducing communication costs
  4. Team member consensus on the idea and a key man to push through
  5. Time, resources, and conditions that allow all team members to focus entirely on the startup

And beyond these 5 things, the most important thing seemed to be “who you do it with.” I learned that team performance varies greatly depending on the combination of people working and gathering together. Willpower alone is not enough. And I had taken startups too lightly.

💪 Before & After

📊 Product Review Analyzer: OpenAI Model Fine-Tuning and Endpoint Development

After the public opinion analyzer project, during my first-ever overseas trip and nomad experience, I built a product review analyzer.

What I learned was: At this point, GPT-4o-mini’s performance in analyzing text and performing multi-label classification for given labels was not that high. For reviews that were slightly more complex or meaningless, the analysis performance fell somewhat short of production-level utility.

Because there were so many problems with the client’s data, I had to manually perform human labeling for the train and test sets one by one.

💪 Before & After

🛩️ First Overseas Trip and Digital Nomad

I went on my very first overseas and long-term trip, lasting two months. I am truly grateful to Professor for his research funding, which made this experience possible. The journey went Tokyo -> Bangkok -> Bali -> Kuala Lumpur. Dividing what left a deep impression by aspect:

⭐️ Digital Nomad

  1. Expanding the breadth of opportunities and perspective when shifting focus from Korea to the global stage
  2. The situations you can think of are limited to your immediate environment unless you are additionally exposed to materials through books or the internet
  3. The ultimate English training was being placed in a situation surrounded by foreigners where you had no choice but to speak
  4. What matters most to me when working is face-to-face interaction with teammates

⭐️ Overseas Travel

  1. You see only as much as you know. This greatly affects the fun factor of travel.
  2. To do that, you need to know yourself well. (What you like, what you want to dig into)
  3. Going to see natural landscapes you cannot find in Korea is the GOAT. (Nusa Penida Diamond Beach)
  4. Trying various foods was also a major source of fun for me.

💪 Before & After

Quarter 3

🤬 Amazon Review Crawler & Analyzer

I tried building my own review analyzer. Having experienced OpenAI fine-tuning model serving with FastAPI, I wanted to expand on that experience and build something directly. As a toy project, I built a review analyzer that can automatically analyze a large volume of Amazon reviews. I finished developing the Amazon review crawler, prompt engineering for the model, and an endpoint for serving. If developed further in the future, I would like to create a sharp service that analyzes and visualizes a large amount of video-based reviews for specific products. Getting video-based information about a product currently takes way too long.

📰 Public Opinion Analyzer Project Resumed

This was the stage where I developed an execution file for a POC and thought about meaningful visualizations in Looker Studio. I participated up to the POC level, after which my mentor T-hyung took over, and I moved on to another project.

For development based on problem identification, problem-solving chains, and ideas to be fun:

  1. There needs to be a consensus on time, place, topic, and who you are doing it with.
  2. Working based on face-to-face interaction with your collaborator and sharing energy.
  3. The feeling of making something of your own, and developing ideas and solutions into decision-making based on communication with yourself or your team. It seems these things are necessary.

💪 Before & After

🏅 CSAT Korean LLM Benchmark Leaderboard

While thinking about quitting the public opinion analyzer, model compression god Hyeong and Jeff suggested developing a CSAT benchmark. Feeling that corporate projects weren’t the right fit for me and wanting to build something of my own, it was a sweet proposal. Following last year’s project, “Scoring Grade 1 on the CSAT Using GPT-4 Prompt Engineering,” I came up with the idea to develop a CSAT Korean benchmark leaderboard for domestic and international models. Giving ourselves two weeks, I automated the model leaderboard for 10 years of CSAT Korean datasets and promoted it domestically and internationally. On the day of the CSAT, I created the 2025 CSAT Korean dataset, ran the benchmarks, and promoted it in Korea once more.

As a result, it became the most runaway successful project I’ve ever experienced. For the first time in my life, countless people took an interest in a project I built myself, people created information based on my project, and watching it grow increasingly famous through growth hacking was fascinating and sweet. Experiencing things for the first time in my life—such as requests for interviews from news reporters, appearing in media outlets, various offers, and even people offering funding—was a great experience that helped me understand what kind of developer I want to become and what kind of work I want to do. (Thank you so much for the Claude Sonnet donation! Oh, Max!). What I found most fascinating was that usually, when you build something technical, interest is limited to technicians and developers similar to yourself, so the scope of attention isn’t that broad. However, the title “CSAT Korean O1-preview Achieves 97 Points” was quite a stimulating topic in Korea, so a lot of people knowing about the project I worked on was a special kind of fun.

I haven’t yet seen a direction in the topic itself where I can drive my technical growth and expand the benchmark. As a plan, I intend to benchmark how much open-source model performance has improved overseas compared to domestic ones in the latter half of this year.

I am deeply grateful to the lab members who lent their strength and funded this project so that it could move forward.

💪 Before & After

Quarter 4

📰 Structured Modeling of Press Survey Statistics

💪 Before & After

🗞️ GeekNews Meetup

I applied for and attended a developer meetup by myself for the first time. It felt completely different from when I went with Jeff earlier this year. I should write this in GeekNews style.

🎸 Miscellaneous

Leaving these down just as things I did since they aren’t of major importance.

Establishing Routines and Second Brain

Running Crew

English Board Game Club

Meetup English Networking

Cooking Club

One-day Class

Dating Life - Blind Dates 5, Group Dates 1

🥳 Realizations of the Year

Better together than alone; if doing it together, the person you do it with matters

The time and energy spent face-to-face with your collaborator are important.

If ideas don’t come, pick a theme and just start

There is no such thing as a perfect idea. Let’s just start.

When receiving information or communicating, grasp the core

My weakness was trying to remember every single peripheral detail, which blurred the core. This is a bad habit of mine carried over from the past. Core first, then peripherals through repetition.

You only know by trying. Let’s think while running

I started viewing work from the perspective of how to finish rather than what to start. The hardest part when reading, communicating with people, or working on something new and unfamiliar was my weakness of not asking out of fear of not knowing, and trying to memorize everything. Moreover, by taking down every single note without knowing what was important, or interpreting things my own way, I would flounder and end up repeating work two or three times. I was trying to solve inefficiency with blind diligence. Or, not really knowing the core of the conversation because it was difficult, I acted like a troll by asking the same questions repeatedly. There were two most important lessons: first, always keep the core in mind, and second, if you don’t know, ask until you know.

Not looking for or asking for a faster, more efficient, and hacky path, and instead proceeding with inefficient methods under the excuse of blind diligence

When rushed for time, it’s hard to produce good code. (For me)

Need practice saying no

Lack of open-mindedness

I tended to lack open-mindedness. There was a confusion between having a blurred subjectivity—listening to others’ opinions and accepting them well—and being open-minded. In fields I don’t know well, my subjectivity tends to blur; how should I solve this? The realization: as soon as you don’t know, ask until you know, and look it up. Ask and research until you are certain, as much as you possibly can. How do you know whether you know or don’t know about something you’re unfamiliar with? In my situation, first…

Helping others is more advantageous; the perspective of finding win-win points with what I can do

There is a time lag before deliverables shine

Just like the CSAT benchmark, the results don’t always shine immediately; you wait a bit until the right timing aligns.

I gained the confidence that even if I’m not great at gathering people, leading, and driving initiatives, I’ll just try it anyway. Once you try it, you realize starting is no big deal.

If you think of anything, you have to start within 5 minutes, otherwise you will procrastinate.

There is no such thing as perfect preparation; you have to think while running.

Regular retrospectives and journaling

Making book reading a habit

I developed my own criteria for selecting books. I choose and read books as a process of resolving dilemmas arising in life. Like a mind map, keyword dilemmas about my life branch out in my head, and books act as nodes to resolve them. Then, books branch out further into new dilemmas that arise after reading, and I pick up another book to solve those dilemmas. Second, I pick up books not for methodologies and prescriptive instructions, but to learn how to think. Most self-help books gently stimulate people’s deficiencies with clever rhetoric presenting their own methods and worldviews to offer solutions. However, they lack substance, or tell stories so obvious that they fail to solve my problems, or I’ve lost direction following them strictly more than once or twice. The biggest problem is that the method suggested by someone is the most suitable method for that author’s life, decision-making process, and circumstances. Therefore, just as everyone’s lived experiences and circumstances differ, the author’s method often fails to greatly help my self-development. Ultimately, one must selectively choose and adapt the author’s thoughts and methodologies to fit one’s own situation. This was inspired by the idea in the book Uncopierable, which states that large enterprises cannot beat small companies that have built ideas and know-how through continuous problem-solving simply by copying ideas. Because of this, I decided to be wary of books where authors merely present methods. Afterward, I view and select books from the perspective of looking at people’s stories and learning how they think, like As a Man Thinketh or The Almanack of Naval Ravikant. I think about: What kind of mindset does this person use to contemplate and make choices? What are their criteria for decision-making? What I felt while reading books

You must start even before fear arises

Decide within 10 minutes. If fear is large, you start pondering whether to do it or not and look for reasons why it won’t work. Since there are countless reasons why it won’t work, let’s just start and think later. Don’t spend too much time on decisions.

Being rich isn’t everything. Is being rich really what I want?

There’s this self-help cliché that you must become rich. My approach, however, was to first raise my own market value. Then, I wanted to do what I love, figure out what I like and what I’m good at, and earn just enough money to get by along the way. If there is something you’re good at, and if it’s aligned with your passion (making a hobby your career), money will naturally follow. So, how much is “just enough money”? There is no problem in this world that cannot be solved. There is only a shortage of time for contemplation and resources.

Shortcomings in ability or unsolved problems stem from a lack of thinking time and brain resources. It is never because of low intelligence.

To solve these problems, everyone has their own mindset, allocated brain resources, and experiences. In the past, I used to think of software development as the realm of geniuses. However, due to constraints and responsibilities in daily life (work, studies), I simply didn’t have enough time to ponder, and my brain resources weren’t sufficiently allocated to solving the problem. This can be addressed through daily retrospectives, and securing that time for retrospectives can also be solved systematically. Ultimately, the reason I fail at something is often a system issue rather than just a lack of understanding.

Is it a world where we can survive just by coding?

I don’t think we can survive by coding alone. The barrier to entry for development is gradually lowering, and unless it’s high-tech, truly core technology, the technology people will use won’t need insane performance—just adequate performance.

The necessity of exercise

I felt this a lot comparing my first semester without exercise to my second semester with exercise. I learned that exercise has a massive influence on guiding one’s train of thought in a positive direction. A healthy mind resides in a healthy body. Mental resilience increases, and when your thoughts become too cluttered, you can sort them out.

If a solution doesn’t pop into your head right away, leave it to your subconscious while you sleep.

Finding what you are good at

The important thing is a mindset of succeeding together, not just by yourself

This is more beneficial in the long run. Among game theory scenarios, the most efficient choice is for cooperation to come first, and if the other person betrays you, betraying them in return is the most advantageous strategy in the game.

The need for a skilled mentor who can stick by your side

The definition and necessity of rest

Scrolling through Reels and Shorts isn’t real rest. Even after resting, it doesn’t feel like a rest. The definition of rest I’ve found so far is continuously immersing oneself in something one finds genuinely fun.

Honesty is not optional, but essential when working

To grow, I must continuously speak about my thoughts and what I know, and be honest. To be that honest, I need to constantly observe and contemplate myself.

Working all day and pulling all-nighters is inefficient

You can’t produce good code that way. On the contrary, if you work for just 4 hours and “must” play for the rest of the time, your productivity increases because you need to finish up to go play, making it sustainable. When tasks are assigned, they aren’t given with the intention that you pull all-nighters, and that should be the underlying intent.

Sheer effort alone doesn’t solve everything. Let’s build a system that enables repetitive habits.

During a book club meeting, the topic that came up with the hardest difficulty was how hard it is to maintain consistency, effort, and habits. In past mentoring sessions and discussions with acquaintances, people often talk about how hard it is to live diligently. Conversely, I didn’t understand this because I used to work and study ignorantly without rest, increasing inefficiency no matter what. This later turned out to be poison for me. It was just fake busyness. You can do things while resting enough. No, you must do them while taking breaks and resting. Instead, it needs to be tailored to each person’s habits, tendencies, and circumstances.

There is no such thing as a perfect choice. Make a choice and make that choice the best one.

This became my best realization of the year, which happened during a meal with Woong-i hyung and Dr. Cheol at the beginning of the semester.

A succession of problem-solving

Life is a succession of problem-solving. To solve problems, how can we ask the right questions?

I need to write periodically. And those writings shouldn’t be based on something I’ve prepared elsewhere, but should organize thoughts that pop up and are observed right now.

Break down tasks as much as possible to create small achievements

This is where I felt the importance of retrospectives and journaling the most. I always made plans and failed to meet them within the set time. I used to brush this off, attributing it to a lack of skill, retreating into lowered self-confidence. However, as I started writing a journal and looking back on my days, I realized that truly, small achievements are crucial.

A person cannot handle more than two big tasks in a day.

I feel like I always set unreasonable goals. Human cognitive limits exist, yet I used to set at least two major tasks like development concurrently and often ended up chasing two hares and catching neither. If I delayed retrospectives under the excuse of being busy, I would end up piling more than two capability-exceeding tasks onto a single day. Consequently, due to the Zeigarnik effect in my brain, I couldn’t concentrate on a single thing, bouncing back and forth without being able to focus on one task in the allotted time, ultimately ending the day having accomplished nothing.

We need a way to speak our minds unfiltered, even if we might be wrong

I realized this through exactly two incidents. First, communicating with Dr. Cheol; second, an interview with a reporter. I need self-confidence. If I don’t know something, let’s honestly admit I don’t know. Or, it’s necessary to confidently blurt out something wrong and get it corrected. That way, a mentor can catch the flawed parts or necessary points to provide direction and correction for the task at hand. In the second incident, the interview with a reporter, strangely enough, despite it being something I created myself, I couldn’t explain everything in front of the reporter. When I explain an idea—even if it’s a 100-value idea that I’ve actually researched and pondered to a level of 100—in front of others, thoughts about potential counterarguments mix in out of fear of being wrong, causing cognitive overload. As a result, I end up sharing only 10 or even negative amounts of information. In my situation, I need to speak logically, confidently, and with grounds, treating what I say as the right answer without filtering it through my brain. Otherwise, the outcome of my achievements becomes negative. Taking the plunge and trying things out is what matters.

My maximum daily concentration span is 4 hours.

This assumes I wake up, read a book to wake up my brain, break down clear tasks within available time resources, and have a clear goal. If this assumption isn’t met, considerable inefficiency arises, or I fail to complete the goal and lose focus on a single task due to the Zeigarnik effect of competing tasks.

If a decision cannot be resolved within 3 days among choices, it means it’s not the right one.

Blocking dopamine

Instagram is completely useless. I installed it for romantic pursuits, but it turns out to be poison. Watching other people’s stories is fun, and posting my own is somewhat fun too, but looking at the long-term, it does more harm than good.

Based on my worldview, maintaining an attitude of searching until I understand if there is lacking knowledge. Also, maintaining a cautious stance against ambiguous or baseless rumors in daily life.

This relates to when MBTI was trending as I entered college. Even now, the realization I got after reading books is that as MBTI trended more, I unknowingly lost my identity, and my worldview drifted according to MBTI types without knowing what I actually liked. It traps you within that framework of thought. And the more you associate with that group, the worse it gets. Moreover, regarding events happening in life, instead of seeking the real reasons, people escape from the real reasons or fears by relying on pre-existing elements like MBTI, fortune-telling, or prophecies, interpreting phenomena through these superficial things and narrowing their thinking.

The language I use reflects my world.

I came to think that ‘language’ in any form is truly important. While working and continuously interacting with diverse people, I noticed that even when conveying the exact same intent, the language people used varied wildly. A person’s personality and actions stem from their worldview, and that worldview is expressed through the language they use. Language is important. I felt this in two instances. First, when explaining very difficult concepts, non-existent concepts, ideas, or abstract notions at work, the speaker and the listener had vastly different thoughts and understandings, and aligning those thoughts took an immense amount of energy. Even during a liberal arts entrepreneurship class, understandings of tasks differed. I also saw this when a researcher hyung and a doctor communicated about mathematically modeling COT—their differing levels of knowledge and experience led to wildly different levels of comprehension. Second, I felt this when meeting random people through networking. My perspective on people changed before and after reading books. When discussing the exact same phenomenon or topic, the language people used was drastically different. I felt this particularly a lot during romantic pursuits: good and bad were defined by a person’s thoughts and language. When facing something perceived as a failure or difficulty, some people might complain, while for others, it can be interpreted as “Does this mean I can grow even more? Lucky Vicky~ Room for interpretation depends on one’s thoughts, and those thoughts are visible in the language one uses.

If you don’t live the way you think, you’ll end up thinking the way you live.

Things I like, based on my current observations

  1. Growth & leading positions (even if not startup-related) - Developing myself, pulling off things I couldn’t do before, overcoming adversity.
  2. Expanding thoughts to a wider world.
  3. Building factories and systems for complex things.

I realized that having my own agency is one of the most critical elements in a project.

What is agency here? It’s my decision-making power when doing a project. Looking at the parts I can decide on:

  1. Project theme
  2. Project direction
  3. Task distribution and definition
  4. Time Etc. Rather than other people’s advice taking the primary role, I should be the one taking that advice and making the final call. Using games makes it easy to understand. There were 3 reasons why I felt this was important: First is fun. When playing League of Legends, if someone micromanages you constantly in your chosen lane, you eventually lose your own judgment and agency, and you just become a shell, plus seeking additional motivation…

Let’s decide within 10 minutes

The 4 elements of work

When working, four elements are important: when, where, what, and with whom. In the order of importance I consider:

  1. With whom: People who match well, and working together helps with growth.
  2. What: What I want to do.
  3. When: Time zones and deadlines that fit my peak concentration and pace.
  4. Where: An environment where I can concentrate well. If I am to work, I want to work in a place that satisfies these 4 conditions. The problem is, while I want to do what I want to make and do, I don’t really know what that is. In times like this, you just have to grab whatever comes your way and try it out.

Things needed

Although I am still at the starting stage technically, looking toward the future, it is necessary not just to stick to development, but to dive deep into one field and then graft it onto another field later. I need one domain to cultivate expertise in.

Abilities I want to gain

I want to gain the confidence that if I contemplate, search books, look things up, and discuss with others, there is no problem in this world that cannot be solved. The same goes for growth in development. You could also call it having my own stubborn conviction (jutaek). I want to have an open mind and solve problems. I certainly like making things. But…

Changed perspectives on money

  1. I observed that if you chase money, aiming for a sudden jackpot, it drifts further away.
  2. Money follows you. Habits that create good luck emerge.
  3. Instead of trying to save money, I should think about how to spend it better.

Opportunities come from people. Shutting yourself up in a corner is not good.

Previously, I thought solitary confinement in a room was the answer. I believed acquiring abilities was entirely one’s own responsibility, but I confirmed that even the environment and paths to acquire abilities come through opportunities from people. Moreover, to be more adventurous and expand a broader worldview, people are necessary.

For me, was starting a business a mistaken belief that it was something I liked? Or was it actually something I admired?

While there are various forms of entrepreneurship, observing the process close up showed me that it repeatedly required going against my own nature, and that achieving explosive growth and rapid execution demanded investing almost all of my time. I also saw how much struggle came from not being able to focus solely on what one wanted to do. Eventually, I watched that startup team halt their challenge, realize their affinity for development and engineering, and prioritize what they genuinely loved over immediate financial gains. After grinding myself down and pushing through to complete a project under such conditions, I asked myself: What on earth am I living this hard for? Thinking back to my original intentions, I had a thirst for growth, and money wasn’t the driving factor. I was simply overcoming adversity for the sake of future endeavors. Let’s just try it first and think about it later.

Life doesn’t go according to plan. Let’s build habits that generate good serendipity

I didn’t expect a leaderboard to appear. Doing text mining, trying out visualization, and meeting the professor and my current lab colleagues were all matters of chance. What I can control is making choices that invite good fortune. To make better choices, I’m thinking about how to cultivate habits that attract good luck. I need to be ready to seize an opportunity when it arises. That is my plan for the new year of 2025.

What does it mean to do a good job? I want to ask this fundamental question.

How to think simply instead of complexly

Everyone’s situation is different. I need to find the approach that suits me.

Everyone’s growth process and path to success are different. Whether it’s developing technical skills, advancing my career, or growing internally, there is a method that fits me. If I assumed I had to follow the exact same path simply because it works well for others, I’d be dead wrong. It’s the same logic as the “building innovation” strategy. Just as small companies tackling a chain of problems soon see others fail if they merely copy solutions while the small startups grow stronger, so it is here.

⭐️ Achievements

  1. Development Skills
  1. AI
  1. Meaningful Visualization Skills
  1. Text Mining
  1. Identifying Weaknesses
  1. Key Realizations
  1. Projects

🤗 Areas Lacking

Watching English content without subtitles took a backseat when romance entered the picture; English networking

As the second half of the semester progressed, I focused more on other things rather than technical growth. I have many regrets about not diving into things like design patterns and architecture.

Need growth in non-development areas; relying solely on development might make it hard to make a living, or I need to do a deep dive into development

Focusing on only one thing, participating in too many projects

Scheduling breaks and learning how to rest with high quality

Somehow, as the year-end got busier:

  1. Routines broke down
  2. Issues with not being able to focus on one thing occurred, I think. Looking at both the first and second halves of the year, I failed to concentrate on a single thing.

🌎 2025 Plans

Top 3 Most Needed

  1. Ability to grasp the core / Preventing the habit of trying to remember everything
  2. Development skills, architecture / Problem-solving ability
  3. Ability to absorb knowledge from a mentor and engage in self-learning

Direction and Themes of the Plan

Themes

Direction

  1. Goal Setting and Execution Strategy

    • Pick good topics and immerse myself in a single project.
    • Improve through rapid execution and repeated failure.
    • Optimize plans and retrospectives on a weekly, monthly, and daily basis.
    • Focus on a single task to foster technical growth.
    • Whereas last year involved multiple projects, this year I will focus on a single theme.

  2. Things to Make and Topics

    • Solve problems and voluntarily complete a full-fledged project or product.
    • Find heart-throbbing topics → try various domains.
    • Expand the leaderboard: Build a representative leaderboard for Korea.
    • A project incorporating reasoning processes and models into LLMs.
    • I want to immerse myself in a challenging endeavor. I want to experience stubbornly tackling a single problem until it’s solved. To do that, I need to grab onto one thing, focus, and effectively filter out noise from other topics.

  3. Habits and Systems

    • Form habits that create good serendipity.
    • Secure proactive and sustainable rest and writing time.
    • The habit of summarizing and condensing notes.

  4. Activities I Want to Execute

    • Complete 1 project per week.
    • Binge-watch Transit Love, decorate a diary, do Pilates, grow a beard, and try various other things.
    • Reform the book club (to satisfy both J and P types).
    • Weekly paper review with Dong-geon.

  5. Core Direction

    • Stubbornly immerse myself in and solve a single challenging problem.
    • Translate abstract thoughts into concrete, actionable goals.
    • Focus on making “fun and meaningful things.”

I want this to be a year where I concretize my abstract thoughts about myself, my abstract pursuits, and the things I like. Even saying this sounds abstract.

공개 여정 구독

글이 올라오면 이메일로 보내드릴게요.

개발, 창업, 배움, 실패와 회고를 짧고 솔직하게 기록합니다. 검색으로 스쳐가는 독자가 아니라 여정을 따라오는 팬을 만들기 위한 뉴스레터입니다.

뉴스레터는 준비 중입니다. 그동안 RSS 로 구독하거나 이메일 로 알림을 요청해 주세요.

Share this post:
Share this post on X Share this post on Facebook Share this post on Telegram Share this post via email

Comments

Sign in with GitHub to leave a comment.

Previous Post
If a Problem Only Tells You That A and B Are Dependent, Without Giving the Intersection Probability or Conditional Probability
Next Post
Overfitting with k-fold cross-validation