DEV Community: 転職カメ

¥10M Job Challenge - Stop the countdown

転職カメ — Sat, 07 Jun 2025 14:17:19 +0000

Due to the nature of my work, I rarely have personal time. Even though I’ve been forcing myself to write a daily journal for over 256 days, as an engineer, I honestly don’t think it’s an effective approach.

Sometimes it takes me an hour just to plan and write, and the worst part is it distracts me from more important tasks or gets interrupted. So from now on, I’ll be focusing on output that actually delivers results.

I won’t give up on the day I reach my goal—what matters is not letting this become a meaningless routine.

Instead of keeping a daily journal, I’ll track how much time I spend on each task—that’s the kind of record that actually makes sense.

Insanity is doing the same thing over and over again and expecting different results.

235/365 | ¥10M Job Challenge - AI Limitations

転職カメ — Fri, 06 Jun 2025 14:32:24 +0000

Today I ran into a textbook example of a problem that AI just couldn’t solve. I ended up spending almost an entire day digging through the database and eventually found out that the issue wasn’t just related to some unique key conflicts—it was also due to the database field types being insufficient to store the data, which meant there were no logs recorded.

In other words, when more than one variable is involved, it becomes very difficult for AI to pinpoint the issue. I even asked it to suggest "possible causes", but it still couldn’t help—it didn’t mention either of the problems I found.

Of course, when you're dealing with issues caused by large volumes of data, directly connecting AI to the database over SCP is far too risky—especially since this particular problem only occurred in the production environment, not in staging or dev. So clearly, that’s not the right approach. Hopefully, no one ever thinks it’s a good idea to hook AI up to a nuclear weapon system...

Another challenge is when there are multiple variables at play, and the project you’ve inherited already has tons of issues—ones that just weren’t caught during client acceptance testing. You only realize them when you dig into the code. So when you try to fix one bug but end up uncovering two or three more, AI also struggles to help in that situation.

At the end of the day, no matter how far AI advances, we need to build a solid foundation ourselves if we want to maintain our edge.

234/365 | ¥10M Job Challenge - Almost gave up...

転職カメ — Wed, 04 Jun 2025 12:47:52 +0000

I suddenly couldn’t publish posts and kept getting errors. I started wondering if I had triggered some kind of content moderation again, it reminded me of some bad experiences on other social media platform, and I was close to giving up. But today, it seems to be working normally again.

Of course, I ran multiple tests myself to confirm that the issue was only affecting my account. I’m guessing one of the dev.to engineers saw the error logs and fixed it? I tried so many things without success and was about to give up, but I didn’t want it to end like that...

As an engineer, even though I’m confident I can reproduce and clearly describe the issue, I still prefer not to contact support—back-and-forth takes too much time, and you never know when (or if) it’ll be resolved. Based on my previous experience on other platforms, they never replied and didn’t even allow follow-up messages...

233/365 | ¥10M Job Challenge - Not that fast

転職カメ — Wed, 04 Jun 2025 09:57:32 +0000

We talked yesterday about how defining requirements often takes a team effort and a certain amount of time before things become clear. Even then, it’s hard to guarantee that the logic is 100% flawless.

Another point is that in large-scale projects, we usually only have access to part of the codebase. (Due to security reasons, access to other systems is restricted)

In scenarios where multiple systems are running in parallel, AI can be quite fragile. In contrast, humans can rely on experience to make educated guesses and solve issues through effective communication—which ties back to what I mentioned earlier: human-to-human communication is hard to replace.

I run into infinite loop issues with AI almost every day. Even with clear prompts, this still happens from time to time. Sometimes it feels like I spend more time crafting the prompt than I would have spent developing something without AI—like teaching a child who never really learns...

Of course, as an engineer, I believe these problems will eventually be solved. I just don’t think it’ll happen as quickly as the media hype would have us believe.

232/365 | ¥10M Job Challenge - Perfect Requirement Specification?

転職カメ — Mon, 02 Jun 2025 14:35:25 +0000

Let’s say that at some point we have AGI or even ASI. I’ve noticed that while many experts believe it could happen within 5 years—and that we might no longer need engineers by then—I still have one question: is there any chance that within those 5 years, both PMs and clients will suddenly become capable of clearly, completely, and logically defining requirements?

Anyone with real engineering experience knows that even with years of design work under your belt, it’s rarely possible to determine the perfect implementation approach in one go. So even if we have ASI 5 years from now, does that mean no one will need to define requirements anymore? I doubt it. Requirements often have flaws, and most of the time it’s the engineers who raise concerns and help users understand when a requirement just isn’t feasible.

Anyone who’s been involved in end-to-end design and development and has around 3–5 years of experience probably gets what I mean. But right now, the market is flooded with fear. Personally, I respect and look forward to the development of AI. As for the fear-mongering, I just treat it like advertising.

Perfect requirement descriptions usually come from collaboration among multiple stakeholders. Even when we implement exactly what the client asks for—knowing full well there may be logical issues or limitations—it often still needs changes once it’s put into real use. Can writing out requirements alone really solve everything?

231/365 | ¥10M Job Challenge - From vague to clear

転職カメ — Sun, 01 Jun 2025 07:57:33 +0000

Has AI changed our software development approach to a top-down model?
In my opinion, it really depends on the company's original culture. Personally, I’ve experienced both top-down and bottom-up development approaches. I’d say the current AI-driven development process feels more like a transition from uncertainty to clarity. In other words, the stronger someone's technical foundation, the better they can validate AI-generated code.

Letting AI handle everything (a.k.a. vibe coding) will likely lead to unmaintainable code in the long run. I don’t deny that some people use this method to release quick-and-dirty products and even manage to profit from them, and I’m honestly happy for them. But it’s just not a viable approach for my work.

So when junior engineers started to disappear, the main competition shifted to mid-level engineers. If at that point you chose not to grow into a senior role but instead fully relied on AI without truly understanding the implementation, it just doesn’t seem realistic.

In fact, the baseline for mid-level engineers is rising—they’re expanding their expertise. The real difference lies in whether someone is willing to properly leverage the tools available. In that sense, things aren’t all that different from before.

230/365 | ¥10M Job Challenge - Interview Preparation

転職カメ — Sat, 31 May 2025 18:15:42 +0000

I recently updated and streamlined my resume based on my recent work experience.

The longer your career history, the harder it actually becomes to write a strong resume or perform well in interviews. It’s easy for there to be gaps between what you’ve done and what’s expected. But at the very least, I want to clearly communicate my experience — essentially, to market myself. So I put together a few key points for interviews:

A concise and impactful resume
Be able to hold a conversation — do some research on the company and prepare a few topics to help ease interview tension
Understand both your strengths and weaknesses (highlight strengths without sounding arrogant, and explain what actions you've taken to improve your weaknesses)
Make sure your traits align with the role you're applying for

No matter what you're trying to learn, it always takes time. But before anything else, it's important to know how to express yourself well, and anticipate what questions might come up based on your experience.

I always stick to the STAR method:

Situation
Task
Action
Result

I tend to get nervous and have trouble showing my full potential, so I keep reminding myself that interviews go both ways — companies are also looking for the right person to do the job. At worst, I treat each one as practice. Confidence takes practice; it's not just about telling yourself to “be confident.”

With a busy schedule, it often feels like my goals are far away, but I believe I’m gradually getting closer. Never give up!

229/365 | ¥10M Job Challenge - Lifestyle

転職カメ — Fri, 30 May 2025 20:41:01 +0000

My usual night owl routine has turned into an early-to-bed, early-to-rise lifestyle. That said, I do feel much more refreshed in the mornings and ready to take on a new weekend.

Lately, I’ve had a bit more free time, so I’ve been practicing some coding problems using Python. But with modern AI easily solving these challenges, I’m not planning to focus too heavily on LeetCode. The main goal is just to get comfortable using Python as my primary language in interviews—getting used to the syntax and commonly used APIs.

Let’s go—here’s to a beautiful weekend!

228/365 | ¥10M Job Challenge - Ability to judge and choose

転職カメ — Thu, 29 May 2025 10:31:12 +0000

In the world ahead, what we may need more than anything is the ability to judge whether something is real or fake, evaluate the quality of AI tools, and determine which ones are suitable and which should be discarded.

In other words, it's about having the ability to make sound judgments and smart choices—something AI can't really give us. It might offer some hints, but the final decisions still rest with us.

Since ChatGPT was released in December 2022, people have been saying that jobs are doomed. But here we are in May 2025, and I'm still working just fine. In fact, I've seen even more new job opportunities and future directions emerge.

Of course, there are still many who insist that jobs will be replaced, trying to stir up panic to grab attention. But whether they're right or wrong doesn't really matter—the real question is: what decisions and choices will we make going forward?

227/365 | ¥10M Job Challenge - Output

転職カメ — Wed, 28 May 2025 17:11:34 +0000

Before I started writing blogs, I didn’t realize that just thinking about what to write actually takes up a lot of mental energy and can seriously distract me. This can create a lot of pressure.

That’s why later on, I decided to just say whatever comes to mind, and not force myself to produce high-quality content every time.

The same goes for work and career transitions. If you don’t see any output, it’s hard to succeed. This is actually something I’m lacking the most. To reach this goal, I think the key is to let go of perfectionism if you want to keep going.

I believed that one of the most important traits for success is the ability to "persist in output."

So at least I’ll stick to this habit. Let’s keep going!

226/365 | ¥10M Job Challenge - Short-sighted?

転職カメ — Tue, 27 May 2025 17:22:03 +0000

Everyone knows the importance of mid- to long-term planning, but in reality, we rarely get to choose how much time to spend on short-term goals—what’s right in front of us. That doesn’t necessarily mean we’re being short-sighted.

Although it’s not something you see discussed much online, I often hear people around me criticize those who are struggling financially, saying they’re just not working hard enough. Maybe that’s true to some extent, but when survival becomes the priority, it’s really hard to think about anything that’s not right in front of you.

In other words, in that kind of environment, your perspective naturally becomes narrow. Even if you’re aware that you’re missing the bigger picture, focusing on immediate needs already takes everything you've got—so it's not that people don’t want to think long-term, it’s that they simply can’t.

So how do we deal with that? My approach is to always set boundaries between work and personal time—no matter what. Like I’ve said before, this is easier said than done, especially in the IT industry where you often need to use your personal time to pick up new skills for work. I’m still in that boat myself. Starting this journal is actually a way to remind myself not to lose sight of mid-term and long-term goals.

Of course, I’ve been learning through trial and error. Recently, due to some changes in my role and my health and focus gradually recovering (it took almost a year), I’ve decided to make a few changes:

Put health first(sleep & diet)
No work on weekends unless absolutely necessary.
The goal on weekdays is to make sure my weekends are truly my own.

I need to carve out time to plan the upcoming week. Since unexpected things come up all the time, planning by the day just doesn’t work for me—it has to be done on a weekly basis.

My weekday goal, Monday through Friday, is to clear my weekend schedule. I don’t mind working late on weekdays if that means I get both weekend days free—that’s the trade-off I’m aiming for.

How do you plan your mid-term and long-term goals without getting completely derailed by short-term demands?

225/365 | ¥10M Job Challenge - A more efficient way of solving problem

転職カメ — Mon, 26 May 2025 16:53:13 +0000

Be persistent in confirming the issue with the client.
If there are no concurrency issues with the system, directly verify the issue using production data when necessary.
Use the client's account and go through each step and detail that triggers the issue with them.

These days, AI can grasp the overall structure and flow of a project much faster than humans—and with fewer mistakes. That’s what I find most useful about AI!

In most cases, customers don’t mind answering questions—unless they’re unsure themselves, in which case you’ll need to investigate on your own.

Lately, we’ve found that unclear descriptions from customers are often the root of the problem. By following up with more questions, we’re usually able to clarify the issue and resolve it more efficiently.

Ultimately, don’t worry about making the customer feel like our technical knowledge is lacking. If both sides share the same goal, this approach is definitely more effective.