What I Learned Interviewing for My Job Search

TL;DR

Applying within 24 hours of a posting going up seemed to matter more than optimizing any single application. A consistent daily habit beat a handful of "perfect" submissions.
Two questions showed up in almost every recruiter screen — "tell me about yourself" and "why are you looking now" — and they're worth over-preparing, not winging.
There's no universal interview pattern. For the senior ML/data roles I was targeting, system design and ML fundamentals mattered more than LeetCode-style coding, but that's specific to what I was interviewing for, not a rule.
Behavioral prep is easy to under-invest in as a technical person. For anything senior, it's not optional — have real STAR stories ready, not a generic definition of leadership.
The biggest lever wasn't more knowledge, it was reps. My early interviews were rusty; a month or so in, the same material came out a lot more naturally.
Don't overload your calendar. I hit a stretch of far more interviews in one week than I could properly prepare for, and it wasn't sustainable — a couple of well-prepared ones beat a pile of rushed ones.

Starting this week I will start a new job in the personalization space, and wanted to write down what I learned going through the interview process.

One caveat up front: this is one person's experience, not a formula. A lot of what determines whether you get an offer is outside your control — demand for your specific skill set, the economy, a company's hiring plans, who else applied, and plain luck. What follows is just my 2 cents based on experience.

Apply early, especially on LinkedIn

The habit that worked best for me was almost embarrassingly simple: every day, look for jobs posted in the last 24 hours and apply to the ones that fit.

I can't prove applying early gets you noticed faster, but that was my consistent experience. There's probably a psychological piece to it too — someone who applies within a day of a posting going live reads as actively looking and specifically interested to the recruiter. Consistency beat perfectionism here by a wide margin.

Once recruiters start calling, be proactive

Once the calls started, I got more deliberate about building actual connections. Before a first recruiter call, I'd connect with them on LinkedIn. Once an interview loop was scheduled, I'd look up each interviewer and send a short note — something like looking forward to hearing about their work and the team.

The important part is not treating this as a hack. I actually wanted to know what the team was building, what tools they used, what they were trying to get done that quarter — and going in with that mindset made the conversations feel less transactional. The point isn't "they'll remember I messaged them, therefore hire me." It's just building a little familiarity and showing you're genuinely interested, which interviewers can tell apart during the interview pretty easily.

Nail the two questions every recruiter asks

Two questions came up in something like 95% of my recruiter screens:

"Tell me about yourself."
"Why are you looking for something new right now?"

They sound like throwaway questions, which is exactly why they're worth over-preparing. Have a structured answer, but don't let it sound memorized — and personalize it to your actual career, not a template.

There's no single interview pattern

I don't think there's one blueprint for what an interview loop looks like. It depends on the role, the company, the team's culture, the seniority of the position, and honestly the individual interviewer's own style.

Across my process I got everything from data structures and small class-design exercises to unconventional coding exercises, ML concepts, SQL, and system design. For the senior technical roles I was targeting, the loops leaned much more toward system design and ML fundamentals than classic software-engineering coding questions. But I wouldn't assume that pattern holds for a different level or a different company — it was specific to what I was interviewing for.

Get the fundamentals down before grinding random questions

One mistake I'd avoid: jumping straight into a big pile of random practice questions. Before doing that, make sure you can comfortably handle the common questions for that specific company, easy-to-medium conceptual questions, the core concepts tied directly to the job description, and questions about your own past work.

For the roles I was going after, my prep time was best spent in roughly this order: system design, ML concepts, data analytics/SQL, then data structures and algorithms. That's not a universal ranking — an entry-level role would probably flip it — but it's where I actually got value for experienced ML/data science interviews.

Be ready to talk about what isn't on your resume

One interviewer asked me to talk about something I was proud of that wasn't on my resume at all, and I genuinely loved that question.

Resumes are optimized for deliverables and impact, which is useful but doesn't tell you much about how someone actually feels about their own work. Asking outside the resume gets at what a candidate found interesting, what problems they enjoyed, what they'd do differently, and whether they can talk about their work without leaning on rehearsed bullet points. It's a pretty good filter for who actually cares about the work versus who memorized their own resume.

Go in for a conversation, not a perfect score

This was probably the biggest mindset shift for me. I started out thinking I needed to answer every question correctly, which is a good way to make yourself miserable prior to the interview. What actually helped was reframing it: I'm here to have a conversation and figure out if this is a good fit in both directions.

That's not an argument against preparing hard — preparation matters enormously. But preparation and composure are different things. You can know the material cold and still have a bad interview if you're so focused on the "correct" answer that you get robotic or afraid to say "I don't know." You need to show you're someone they'd want to sit next to for the next few years.

Your first few interviews probably won't go well — that's normal

My earliest interviews in this search were some of my weakest, and it wasn't because I didn't know the material. I was rusty, and still figuring out how to think out loud under pressure. After about a month to six weeks of steady interviewing, something shifted — I got noticeably more comfortable, including in moments where I didn't know the answer.

The change wasn't that I'd learned dramatically more in that window. It was that I'd gotten better at interviewing as a skill, separate from the underlying knowledge. I stopped treating a wrong answer as a failure and started treating it as a chance to show how I think. If your first few interviews go sideways, that's not necessarily a verdict on whether you're qualified — you might just be shaking off the rust.

Prepare seriously for behavioral questions

As a technical person, it's tempting to treat behavioral prep as an afterthought and spend all your time on ML, system design, and SQL instead. I'd push back on that, especially for senior or staff-level roles.

Once your interviewers include people above your prospective manager — a VP, other senior leaders — expect more questions about leadership, collaboration, disagreement, and influence. Those aren't testing technical depth. They're trying to figure out how you actually operate.

STAR is the obvious structure and it works: situation, task, action, result. But don't just memorize four generic answers — think through real moments involving a disagreement with a colleague or stakeholder, a hard technical or business call, a project that went sideways, influencing someone without formal authority, driving alignment across teams, handling tough feedback, juggling competing priorities, or mentoring someone. The goal is to sound like you're describing something that actually happened, not reciting a definition of leadership from a textbook. For senior roles, technical skill is only part of the bar — they're also checking whether you can collaborate, take ownership, and help a team actually ship.

Use ChatGPT and Claude as prep tools, not oracles

I used AI tools a fair amount while prepping, mostly one specific way: feed in the job description plus what I could find on the interviewer — their role, current responsibilities, whatever was on their LinkedIn — and ask what they were likely to ask me.

Honestly, the predictions were wrong more often than not. That wasn't really the point. It gave me a checklist and a starting structure instead of a blank page. A lot of interview stress comes from not knowing the scope of what's coming, and turning that vague uncertainty into something concrete, even an imperfect list, made prep less paralyzing. Half the questions being irrelevant still means the other half made me think about something I'd have otherwise skipped.

Don't overload your calendar

This one I got wrong at first. My instinct was "fail fast, fail often, eventually something lands," so I started stacking interviews — at one point well more than I could actually prepare for in a single week. I don't recommend it.

Each serious interview eats real mental bandwidth, especially once you're prepping specifically for that company and role. Each one had its own flavor — one leaned heavy on modeling depth, another on SQL, another on system design — and constantly switching context between different formats and interviewers burns you out fast. I'd aim for something closer to a handful a week if you're actually preparing for each one properly, and don't let prep eat your current job either. It's easy for interviewing to start feeling like a second full-time job; it shouldn't come at the cost of the one you're still being paid for.

The bigger lesson: prep, repetition, composure

Looking back, there wasn't one trick that got me through this. It was preparation, then repetition, then feedback from real interviews exposing the gaps, then better composure, then better performance — in that order, repeating.

Preparation gets you the material. Repetition gets you comfortable saying it out loud. Actual interviews show you what you missed. Eventually I stopped treating every interview like a test with a pass/fail score and started treating it as a chance to show what I've worked on, how I think, and what kind of colleague I'd be. That shift in mindset mattered as much as any of the technical prep.

My prep checklist, condensed

Before applying

Apply to relevant roles within a day of posting.
Check new postings daily instead of batching.
Don't over-optimize every single application.

After a recruiter call is scheduled

Have a personalized "tell me about yourself" ready.
Have a real answer for "why are you looking."
Connect with the recruiter.

Once the loop is set

Research the company, team, and each interviewer.
Connect with interviewers professionally.
Use an AI tool to sketch likely question areas.
Prep specifically against the job description.

During prep

Start with job-specific and fundamental questions before obscure ones.
For senior ML/data roles, weight system design and ML concepts heavily.
Have real stories ready about your past work, including things off your resume. DO NOT underestimate this.

During the interview

Treat it as a conversation, not a test.
Think out loud, ask clarifying questions, say "I don't know" when true.
Show genuine curiosity about the team and the work.

During the search overall

Cap your weekly interview load — a handful, not a double-digit pileup.
Give yourself room to learn between interviews.
Keep your current job under control.
Trust that interviewing gets easier with reps, even when the early ones feel rough.

None of this guarantees an offer — too much of the outcome sits outside any one person's control. But the parts you can control — preparation, consistency, how you communicate, and how you carry yourself through the process — made a real difference for me.

If you are interviewing, All the best!!