Feeling stuck after 2 years in: skill gaps, AI tools, or wrong fit?

TL;DR
I’ve been working for a little over 2 years across 4 web-dev roles (all remote), regularly logging 50+ hr weeks, but still dealing with low productivity, buggy code, and unclear career fit. I’m seeking advice on whether I’m missing key skills, not leveraging efficient workflows/AI tools properly, or simply in the wrong path.


I come from an engineering background (no CS degree) and pivoted into front-end dev via a 9-month mentorship in 2021. I studied engineering in college so I had a good foundation in programming and the program filled in the gaps. Since then:

• In my first startup, I sometimes received ~60 comments on a single PR while pulling late nights.
• My second startup also involved long hours and ended when it ran out of funding.
• I did a three-month contract role.
• Now at my current position, I work long hours again with minimal review process, and I’m getting feedback about low productivity, frequent bugs, and sometimes poorly written code.

Meanwhile, my peers often finish by 5 PM with seemingly no major issues. I feel that putting in more hours than others would translate into higher quality work and fewer bugs, but this isn’t the case. I use AI tools like ChatGPT and Claude, but they frequently generate subtly flawed code that I must spend extra time debugging. So, I keep wondering if I’m just not using AI tools effectively, or my peers simply have sharper core skills I somehow haven’t mastered yet.


My Questions

1. Skill-Gap Assessment
Given the feedback I’ve received, which core technical or problem-solving weaknesses am I most likely missing right now?

2. Efficiency & Tooling
What workflows, habits, or AI-powered tools have you found truly effective for improving both speed and correctness? How did you adopt them?

3. Career-Fit Signals
In your first 1–2 years, what concrete indicators showed you that web development was (or wasn’t) your long-term career? Did you ever work long hours for long stretches?

4. Prioritizing Growth
If you were in my position, which 1–2 areas (technical or communication-oriented) would you focus on first—and how would you approach that learning?


Thanks in advance for sharing your experiences, advice, or resources!

Note - if you used AI to write any part of this post, please remove an AI generated content, per the forum rules.

It sounds like you are not getting the support you need from your employer. This is the kind of feedback and guidance your team leader and more experienced co-workers should be giving you.

Maybe because you are working for startups?

I would look for any role at a more established company that provides training and guidance.

Kind of answering your own question here.

Thanks so much for replying! That makes a lot of sense. I also have a feeling that the way my career started and the fact that I’ve ever worked only with startups, remotely, might have a lot to do with this. Unfortunately, I’m not in a position to find another employer at the moment. The next best option that I have is to work again with the mentor that helped me switch to tech a few years ago. I’m still concerned, though, that the level of support might not be what you would get when being mentored by a team lead and more senior colleagues. Another issue is I wouldn’t be able to work with the mentor on startup’s code base for confidentiality reasons.

Given this, do you think an external mentor would still be a good be a good substitute?

To be sure, you’re saying that I have skill gaps due to lack of mentoring or working for startups, is that correct?

I agree, that could absolutely be part of it. You miss out on a lot of incidental chat and small questions/advice. Do you communicate frequently in one-on-one or group chats? No pair-programming with a more advanced dev?

Yes. Work with the mentor you had before, or hire a coach or even just a supplemental course with guidance if there’s a particular area you want to work on. Since you are working, you might be focused on working and stopped “training”.

I read about this story, (I wish I could remember the book, maybe something by Malcolm Gladwell) about when Kirk Hammett was hired by Metallica he didn’t stop learning. Even though he was already playing with this big band he hired Joe Satriani for further guitar lessons to get even better.

You do seem focused on improvement already, but the job itself isn’t providing that.

I chat one-on-one, mainly with the manager. I also huddle a lot with him and sometimes hop on office hours for a debugging session. I have to say, though, that I don’t get to code while someone else is watching over my shoulders. It’s usually me observing while the other person codes.

This was different in my first job: heavy PR reviews, pair programming (mostly the other person coding), me asking a bunch of questions on how to do things/fix errors, asking the other person about their thought process. This was before ChatGPT in 2021/early 2022 and I was just starting out, so I couldn’t survive without that level of support. My next job also didn’t have a PR review process, so I worked with the mentor to help me get unstuck and do a bit of code review, until ChatGPT was released.

Now, I don’t get completely stuck on something and don’t need someone to hand-hold me (unless there’s something outside my scope or backed-related). The job isn’t as demanding as the previous one in terms of code quality either. I’m basically asked two things: deliver code that’s bug free on time. For some reason, I find myself working long hours to fulfill this and still end up shipping code that isn’t bug-free. Apparently, my boss isn’t too happy with the bugs (jokes about having to avoid “dead bodies,” though he hasn’t given directly critical feedback yet), and more recently my speed.

You’re saying some contradictory things here. You’re saying you don’t need the mentor support anymore and the job is less demanding yet you use GPT and end up working long hours to deliver bug-free code.

Go back to this. Don’t use GPT.

This is the problem.

I understand why it might seem contradictory. Surprisingly, AI helps me with difficult bugs (those crazy “Syntax Error…”), but can’t give me exactly what I want in terms of styling and functionality, so I end up making a lot of changes. The mentor is more suited to help with a few difficult bugs but cannot help directly with the kind of time consuming grunt work that’s needed to fix up the little issues in AI-generated code. And since I spend most of my time on fixing things, I end up cutting corners because there’s no time for doing proper testing, which means shipping code with bugs.

Thanks for your input though. I’ll see if the mentor can help me figure out a solution.

If you find AI is giving you solutions to hard bugs but with subtle harder bugs in the code, I don’t see how AI is helping you here.

I guess it’s up to you if LLM generated code is costing you time or saving you time.

1 Like

This seems like another big problem with an obvious fix. Don’t spend so much time here (you mean fCC or fCC forum?)

Oh I meant to say I spend most time fixing stuff. I don’t come to FCC that often :slightly_smiling_face:

Ok, well it sounds like you are in an unhealthy relationship with AI.

I think you should rely on it less, only for small sections of easily reviewed code. Kind of like a typing assistant. Maybe also try co-pilot instead of GPT and Claude?

1 Like