Forem: miffens

Debugging “Where to spend my time?” in the job search 🌲

miffens — Thu, 01 Jan 2026 00:22:36 +0000

Something I hear a lot from jobseekers is, “I don’t know where to spend my time.” There is so much you could do—resume tweaks, projects, networking, interview practice—but little signal about which would help most.

Below is the “decision tree” I guide my mentees down, organized by “stuck point.” After identifying where they're stuck, we pinpoint why, then use that to define top priorities.

I don’t know what career direction to take

This “quadrant” could help:

What are you good at, and enjoy?
What are you good at, but don’t enjoy?
What are you passionate about, but not good at?
What are you not passionate about, and not good at?

If you need a job ASAP, choose in-demand roles that want what you’re already good at (1), even if you don’t enjoy it (2). As you gain seniority, you'll unlock more 1 and 3 opportunities. Avoid 4.

Don’t sweat being hyper-specific if you’re entering a new field because entry-level roles are generic. For example, stuck between accessibility guru or design systems specialist? You’ll start as a fullstack generalist anyways.

Should I broaden the roles I’m applying to?

Weigh the potential upside of more opportunities against the extra time required:

Writing a resume version tailored to that role
Learning skills that you’re missing for that role
Preparing for role-specific interview formats

If it overlaps with your existing skillset and interviews you’re doing, it’s often worth trying.

I’m applying but not hearing back

I check for these 3 common reasons:

1. Missing skills/experience

Go line-by-line through a target role and grade each job requirement: My resume...

demonstrates several examples of it
has it but not deeply enough
is missing it

Confirming your grading with industry folks is ideal. Job descriptions are often unrealistic wishlists, and insiders know what actually matters.

This may be your main area to improve if you have 3’s. 2’s should be noted, but prioritize them relative to other gaps. For example, shallow React experience may be lower priority if your resume isn’t finished.

Prioritize adding projects to your resume that address your gaps. If you don’t know what to build, go for “technically robust and universal” over “creative.” Companies prefer candidates who have built features like theirs in their exact stack. Use the most popular tools for the field (React for frontend, Python/pandas for data science, etc.). Focus on features every app has: authentication, forms, data tables, as well as current hot topics like LLM agents.

2. Resume/ LinkedIn not doing you justice

You may be qualified, but your profile isn’t written in a way that fully brings that out.

A common mistake is to submit “general” resumes where all your skills, relevant or not, are given the same space on the page. Instead, have relevant experience take up at least 60% of the page and shrink less related points. For the exact same person, a tailored resume will look like a stronger fit over the general one.

Another mistake is omitting specifics, leaving recruiters unclear about the depth of your skill. Example:

Generic: “Built a chat app with React.”
Better: “Built a chat app (React, Socket.IO, Node) supporting 500 concurrent users. Added OAuth and reduced message latency by 40% by batching events.”

A third is lacking clarity in how an experience relates to the role, particularly those from previous careers:

Unclear: “Managed inventory for grocery store.”
Better: “Managed inventory for 5,000‑SKU store using performance reports and weekly audits. Identified 12% overstock through Excel script.”

For maximizing LinkedIn, I recommend Emily Worden’s content, e.g. How to Write a LinkedIn Headline, How to Effectively Announce ‘Open To Work’.

3. Weak network

If your resume reads well to both you and industry insiders, then the best use of your time could be sourcing quality job leads.

Dedicate time just for finding job sites, forums, or people that specialize in your industry, match your values, are local to you, etc. They can surface less-known, strongly-aligned opportunities. Be active on LinkedIn to boost your visibility. If you see a good role but it closed, you can connect with people who work there and ask them to keep you top of mind when more roles like it open in the future.

I’m hearing back but failing technical screens

Prioritize technical interview prep over projects and networking.

You can further focus your study scope by format:

Leetcode/ algorithms: typical at larger companies, these test your knowledge of data structures.
Hackerrank/ take-homes: also common at larger orgs, these are time-limited tasks asking you to implement a larger function utilizing data structures.
Build a small feature: more common at startups, this incorporates Leetcode Easy concepts in the context of a practical feature. Your “on the job” skills translate more naturally to these.

You can prioritize companies with interview formats that you're stronger at. For example, I’m extraordinarily talented at misreading take-home instructions and implementing the wrong thing. Meanwhile, a human interviewer correcting me before I get started drastically improves my success rate. Therefore, I prioritize interviews at companies without take-homes.

To narrow even more, technical interviews consist of these sub-skills:

Knowing key syntax by heart: print, string/ array manipulation, sort…
Data structures: knowing patterns > completing lots of problems
- Advanced topics such as dynamic programming, heaps, etc. are less frequently asked. Unless you are going for MAANG companies, you can deprioritize technical interview studying once at this point.
Talking out loud: articulating your thought process as you solve the problem

Find resources that work for your learning style to practice your main gaps in isolation, then all together.

I’m fumbling soft-skill / manager interviews

These ask some form of the same questions:

Technical strength/ product impact: “Tell me about your biggest project”
Conflict: “Tell me about a conflict/ disagreement and how you resolved it”
Tradeoffs: “How do you decide what to do when everything feels urgent?”
Cross-functional collaboration: “How do you work with PM’s/ designers?”
Process improvement: “Tell me about a project that failed”
Failure and learning: “What is your weakness?”

Prepare 1-3 specific stories for each in the STAR format, and rehearse them out loud.

Tips:

Writing a strong resume doubles as prep for these interviews!
You can record yourself to emulate performance pressure.
For remote interviews, you can have your prepared answers on-screen—the interviewer can’t see it.

I’m getting to final rounds but not getting offers

This typically means you’re qualified, but other candidates who also made it through are edging you out.

If you’ve made it this far, recruiters are usually willing to give feedback. It’s a win-win to help you and keep in touch, especially if you met their bar but they only had one opening. How I interpret:

If they say something about not meeting their technical bar, work on your technical interview skills
If they don’t, it could be something about your manager screen answers
Some feedback can be unfair/ not a good way to evaluate a “good engineer.” Disregard and be glad for a bullet dodged.

For a more accurate analysis, a hiring manager that hires for your role is the ideal person to get help from at this stage. Show them your recruiter feedback emails, present your manager interview answers to them, etc., and they can pinpoint how you could further improve.

Spacing out final rounds and dedicating time to reflection and improvement can be an overall time-saver. I once scheduled 3 final rounds back-to-back, applying the same approach, and got rejected by them all. After some coaching on my manager screen answers, I received an offer soon after. In retrospect, those 3 rounds felt like needlessly hitting the same wall.

Beyond the “Priority Playbook”

Balance priorities with what’s fun

This guide pinpoints your objective next best step, but that can be the toughest one. Balance it with tasks you have more energy for that also benefit your growth. For example, if you’re a web dev that prefers Vue, go for it—the foundations carry over. If you’re in the early rounds and dread Leetcode but find system design more interesting, do it—you will need to prepare for it down the line anyways. If you enjoy writing, blog about your learning to broaden your network.

Focus on the current step

It’s overwhelming thinking about all the upskilling and interviewing you’ll need to do. However, there’s no need to think about all the steps at once, as long as you know your current steps are must-do’s towards your destination. Like a video game, take it one level at a time.

Be curious, not pressured

"I'm ‘running my code’ to see what knowledge gaps pop up" is more approachable than "I have to nail this interview." Aim to learn, not to pass.

“I’m stuck and making no progress” can be reframed as “I’m stuck because something isn’t working, and that’s valuable information. It helps me eliminate what -not- to do and get closer to what does work.”

Closing thoughts

Your job search can be thought of like a decision tree algorithm: each step is about finding your next best moves, not solving everything at once. Strive for continuous, intentional iteration.

Any questions or want a deeper dive into any of these? I couldn’t possibly fit all my mentoring and resume writing experience in one post. Let me know what you’d find helpful, I'd love to help!

Also available in Bluesky thread format 🧵:

How to get past “I don’t know where to spend my time” in the job search, a thread 🧵
— Shell 🍋 UX Dev/ Resume Writer (@miffens.bsky.social) 2026-01-01T00:01:57.545Z

(Editorial assistance: Claude 3.5 Haiku)

How I learned to “use AI” in 3 baby steps

miffens — Tue, 11 Nov 2025 05:25:13 +0000

In recent conversations, I've heard a lot of anxiety about "AI fluency." Between companies pushing adoption and jobseekers getting interviewed on how they "use AI," coders worry they're falling behind. On the other hand, they’re also uncomfortable drastically changing what works well for them.

If you feel you need to "use AI" but aren't sure how, here’s what I did to go from "intimidated" to "fluent" without getting disrupted.

Step 1: Make it exist in your IDE

Set up Cursor or Copilot and... do nothing special. Ignore the chat for now, only let it suggest completions and docstrings as you type. Sometimes it seems psychic, other times it'll phrase comments more eloquently than you could, and often it's nonsense. Continue going about your usual workflow while occasionally accepting its suggestions.

Congrats! You've taken your first steps towards "AI fluency." At this point, it's really a smarter autocomplete and shouldn't overhaul your existing workflow. I stayed in this stage for about a month.

Step 2: Try inline editing

You know those tiny snags: awkward syntax, a type error, a small refactor where your intent is too nuanced to find a direct Stack Overflow answer for? Hovering over the red squiggly line or highlighting some lines should make a "quick edit"/ "fix with AI" option appear. Give it a go! Accept if it fixes it and otherwise undo. If you're up for it, edit the default prompt to add context or specific instructions to up your chances at a good edit.

From the Cursor inline edit docs. Yes 'unused' is misspelled, maybe to show that LLM's are forgiving of them? 😀

Step 3: Use the chat agent

When I posted codebase questions on Slack, coworkers would reply, "here's what I know, but have you also tried asking Cursor? It's pretty good at answering your type of question." That was my sign from the universe to try the chat agent for bigger stuck moments and legacy spelunking. As I interacted with it, I started to build an intuition for what it does and doesn't do well.

What I ask most:

Fixing - "My code isn't working. I expect X but it's doing Y. Here's the error log."
Explanations - "Help me understand how function X affects rendering. Walk me through concrete examples."

I write to it like a coworker who is helping me, providing all the context on what the problem is and what I've tried so far.

Nonsensical rambles could happen, but often, it will point to the right neighborhood even if it doesn't hand me the final fix. Once in a while, it nails it. I review its edits as if a coworker edited my branch, with about 20% extra caution.

Switching from 'Auto' to the best model and the largest context window my company provides, GPT-5 MAX, led to slower but much better answers. Getting helpful answers more often motivated me to keep experimenting with it.

For a structured course on chat agent skills, I genuinely recommend the free Anthropic AI fluency course (1-2 hours) which emphasizes how to collaborate effectively and ethically with LLM's:

Delegation: Thoughtfully deciding what work to do with AI vs. doing yourself

Description: Communicating clearly with AI systems

Discernment: Evaluating AI outputs and behavior with a critical eye

Diligence: Ensuring you interact with AI responsibly

Conclusion

Now that you have a sense of when to leverage autocompletes, quick edits, and the chat agent, you know how to "use AI" at work. Congrats! With these fundamentals, you should have the confidence to lean further into more advanced usages if you're so inclined.

The learning curve is non-zero, but your Stack Overflow, pair programming, and code tracing strategies both transfer directly and mix-and-match with your new skill. Baby steps will get you there. You got this!

How I got better at e-learning by ignoring the completion bar

miffens — Tue, 11 May 2021 17:03:21 +0000

Intellectually ambitious but with limited spoons, I've experienced the blessing and curse that is the abundance of e-learning resources. With these resources accessible as e-books, video lessons, and interactive exercises that tailor to all levels, it paints the picture that anyone could learn this desired skill, if only they had the discipline to finish it all.

However, in striving to meet all needs, course curricula extend to such a length that the time commitment to complete all of it becomes a hurdle. It's common to bookmark or purchase but never begin, or fade out within the first 10%.

After trial and error, I found that this linear, 100%-completion oriented approach was a red herring goal I needed to unlearn. If I lose steam on a course, it's usually not a lack of discipline, but a signal the current lessons aren't valuable enough to me for the effort I have to invest.

Here are some ways I've responded to that signal:

Skip ahead to lessons that sound more interesting. These may require skipped lessons as a prerequisite. With a better understanding of the other lesson's purpose, I have renewed motivation to work through it.
Know when to prioritize breadth. Full absorption of every lesson through replay and exercises without peeking at the answer might have caused too much friction. If deep understanding isn’t urgent, I'll decide that a single pass is enough on the remainder. In contrast to fizzling out early on over-engagement and being unaware of the other course topics, I can recognize key words and know exactly where to go for deeper understanding.
Know when to prioritize depth. Identifying the 2 out of 10 chapters most relevant to me and deciding the other 8 would give diminishing returns is liberating. Then, I can study those two extra carefully and move on to the next course sooner.
Research a different course/ learning medium. Learning from another resource that offers a more intuitive teaching style for me could save tons of energy.
Identify other higher priorities. Am I unmotivated because there's no real urgency? Is there another more fulfilling place to spend the same time?

These approaches have helped me maintain momentum that I otherwise would have lost, and see self-experimentation an investment to account for in my learning pace expectations.

A Design Challenge for E-Learning Platforms

As an e-learner, my main goal is to upskill through a learning path that works best for me. But what does the typical resource signal with its user interface? Completion is everything. Follow the path. Stay the Course. Get that check mark of success.

Completion bars, next lesson autoplay, disabling future lessons until preequisites are met, make the cherrypicking approach not readily discoverable. Furthermore, even if I know I extracted the most valuable pieces of the course, the unsatisfying misalignment in the sad faded unwatched videos and the happy checked ones make me feel like a poor student.

What would a site that facilitates and rewards a variety of learning styles and goals look like? How could a user interface celebrate nonlinear, cherrypicked learning? Could a workflow guide users through that curation process, or self discovery of one's learning style?