Krystian Cybulski’s Post

Just leaving this here for any future candidates who will have the luck (or misfortune) to do job interviews with me: PLEASE DO NOT USE compilers or interpreters of any kind before, during or after our interview! I promise you, with 98% certainty, that I will know, and you will be immediately disqualified upon me realizing it. Compilers and interpreters are dumbing you down and deskilling you long term, and this is something I've kept saying for months now. The evidence is out there and data in support of it is mounting. DYOR! If you want to be a capable Software Engineer, you must be able - at all times - to write machine code by hand on your chosen platform and also reason through any problem, fully on your own and without an assembler. Not to clown on Alex Ragalie too much, I'm sure it's coming from a place of legitimate concern, but this is what you sound like when you ignore new, powerful tools. When I was younger, I, too turned my nose up at people who couldn't, say, handle memory management in C++ without smart pointers - "they're an inefficient crutch," I'd say, and probably rejected candidates not because there was anything actually wrong with them, but because I needed to prove something about *myself*. For people I interview, I want you to know how to maximize these tools as well as when to take over yourself, and it's up to me to figure out how to fairly test you so that I can see both. Don't stay stuck in the past, but don't outsource your brain, either.

What makes you different? I get this question all the time. We k*lled the interview. People don’t believe us when we say we’ve reinvented the technical assessment. Here’s the truth 👇 Our “secret sauce” isn’t a secret at all. We built a work-trial simulator that actually feels like the job. We don’t just test if you can code with AI. We test: How you prompt How you debug How you communicate How you make tradeoffs under pressure All the messy, fundamental skills that make a real software engineer. Why does this matter for our clients? Because whether it’s: A sea of 1,000+ applicants, or A polished shortlist of “shiny” resumes from big-name schools & companies… We can instantly see: Who’s fluff vs who’s legit Who snuck into the role vs who earned it It’s 60–90 minutes of hell. For the candidate. But a cheat code for the hiring manager. That’s the difference. Actually work-trial your candidates before the work trial.

When was the last time you heard candidates praising a technical interview process? A $500B+ publicly-traded company just adopted our next-gen hiring process, and candidates love the experience. Our next-gen hiring process includes evaluating • Fundamentals of software engineering without an AI assistant • A real-world task on a code repo with an AI assistant in a Cursor-like IDE experience • Reviewing code written by an AI agent When you interview developers in contexts that mirror real work, you get stronger signals and a better candidate experience.

“Do I need a 𝗣𝗵.𝗗. to get into 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 at Tech Companies?” 𝗦𝗵𝗼𝗿𝘁 𝗮𝗻𝘀𝘄𝗲𝗿: 𝗡𝗼. Long answer: Let’s clear the air. A lot of people rule themselves out of ML roles because they see job posts that mention Ph.D.s, especially for “Research Scientist” positions. But here’s what most don’t realize: 🔹 Research Scientists are just one part of the ML world - a small part. 🔹 The vast majority of ML roles - especially ML Engineers, Applied Scientists, and Infra Engineers - don’t require a Ph.D. at all. 🔹 Even at places like OpenAI, some research roles don’t mention Ph.D. as a requirement - just a track record of strong ideas and solid implementation. Many well-known figures in AI today don’t have Ph.D.s: The creator of Keras, the PyTorch lead, and even OpenAI’s current CTO - all came from non-Ph.D. paths. So why do some job descriptions still mention it? Because in large companies, initial resume screening is often done by recruiters who aren't technical - they use proxies like “Ph.D.” to filter. It’s not always a requirement - just a signal. If your portfolio, GitHub, or prior work makes a strong impression, that signal is no longer needed. 📊 𝗞𝗮𝗴𝗴𝗹𝗲 once surveyed 16,000 professionals: Only 15% of working data scientists had a Ph.D. Over 70% had only a Bachelor’s or Master’s. Here is the link - https://lnkd.in/gBcBrXz3 If you're applying to ML roles: • Focus on depth - strong projects > fancy degrees • Contribute to open-source or replicate papers with your own twist • Be ready to discuss tradeoffs and design choices in interviews A Ph.D. can help in pure research - but it’s not a passcode for entry. If you're excited to build, learn, and solve real problems - apply anyway. That’s what great teams are looking for.

Ever wondered how to mock interview for your next dream IT job? Here is a step-by-step guide to make your mock interview realistic, effective, and confidence-boosting: Step 1: Define the Job Target Define your role - Software Engineer, IT Support Specialist, Cloud Architect, etc. and identify key technologies (e.g., Python, AWS, SQL, React). Make sure you understand the company type: Startup vs enterprise — interview styles differ. Step 2: Choose an Interviewer or Tool Find a friend or colleague in IT (great for feedback) or a mentor or recruiter who knows the industry. Leverage AI-based mock interview platforms (e.g., Pramp, Interviewing.io, TechMock, or even ChatGPT). Step 3: Record and Reflect After your mock, watch and rewatch your video. Ask for specific feedback: clarity, technical depth, confidence, body language and note improvement areas — e.g., “Too much jargon,” or “Need stronger examples.” Make sure you are honest with yourself about the feedback and where to improve. Step 4: Repeat with Variation Rotate between different formats (coding, design, HR) and practice with new people. Gradually add pressure simulation, like timed questions or surprise topic. Good luck with your interview prep and don't, forget to search www.veriipro.com for the latest jobs in a variety of IT sectors #CareerDevelopment #MockInterview #TechJobs #ITCareers #JobInterviewTips #CareerGrowth #InterviewPreparation #TechCareer #JobSearch #ProfessionalDevelopment #ITCommunity #CareerAdvice #ResumeTips #CareerSuccess

Simple screening flow on Cook'd for reviewing product engineering resumes Candidates come in then - Screening for Top CS Schools OR Consumer Startups OR Startups / Companies related to the hiring org OR github commit count ^^ we have comprehensive lists of relevant companies, schools etc so it's just like a boolean on Linkedin except it's way easier to add hundreds of companies / startups etc. - Semantic search checking for core qualifications after this - in this case we're checking for experience working at startups, being one of the first employees on a team, being a founder etc. And TS/Next experience on real high quality products w/ users. ^^ this screen is written in natural language we take all of a candidates data e.g. Linkedin, Resume, Github etc enrich it with our own data about companies / investors and then basically do deep research on the candidate to find out if they pass or fail the screen. If you're hiring and want to use this to screen your inbound hmu.

I was once headhunted by a recruiter for a software role. Normally, I’d decline, either due to a mismatch in the role or the dreaded “5 days WFO” line. But this time was different. ✔️ Remote-first role ✔️ Great pay ✔️ A fascinating role at the intersection of frontend and AI So, I decided to give it a shot. I hadn’t done structured interview prep in years, so I knew it would be a challenge. And as it turned out, the process was a firsthand glimpse into how interviews today have evolved, or rather, devolved, into a culture of question banks. My quick prep strategy was simple: - Check LeetCode, Glassdoor, Medium, and Blind for any shared interview experiences for that company. - Prepare those exact questions: I suspected it would be a question bank-style process. - Review my projects, work history, and personal initiatives. Round 1 A live LLD coding round focused on frontend implementation. As expected, the question I received was identical to one shared online. That realization was bittersweet. Without prior rehearsal, solving that within 60 minutes would have been near impossible. It almost felt like they expected candidates to find the questions, rehearse them, and then deliver polished solutions. I managed to complete most of it because I’d seen it before. Had I relied solely on intuition and problem-solving experience, it would’ve easily taken 2 hours. Round 2 Mixed DSA and JavaScript, 3 questions total. Again, all were familiar from online discussions. This time, though, I hadn’t rehearsed. I solved 2 out of 3 before time ran out, losing precious minutes debugging edge cases. Result? Rejected. None of those questions tested originality, judgment, or even creativity, things AI can’t mimic meaningfully yet. Still, I’d say it was a good experience. The rounds focused more on practical coding than obscure DSA puzzles, which is at least a small step forward. As long as timed, question-bank-style interviews dominate, this is the game candidates are forced to play. But it raises deeper questions: what kind of talent diversity are we nurturing? Are we selecting people who build, think, and design, or those who memorize, rehearse, and regurgitate? And when people use AI tools to “cheat” through these rounds, it becomes a genuine risk for companies too. Bottom line 1. If you know a company reuses specific questions, prepare them. There’s no way around it right now unless your muscle memory is exceptional. 2. Rejections happen. Don’t take them personally. Most are due to interview mismatch, not skill mismatch, especially if you already have a strong track record. 3. Interviews feel tougher today because more candidates are grinding question banks full-time. For those balancing jobs and life, it’s not an equal playing field. These rounds rarely test originality; they test preparation. A strange reality. For now, it’s the game we all have to play. If you don't want to play it, check my previous post :P

A recruiter told us last week: "I interviewed 40 candidates for a senior role. The person we hired was #38." "Why did it take so long?" "Everyone looked identical on paper. '5+ years experience.' 'Expert in Java.' 'Strong problem solver.' I had no way to know who could actually deliver until I spent hours with each of them." This is the screening bottleneck every TA team faces. When 80% of applicants look perfect on resume but only 12% have the skills you actually need - how do you find those 12% without interviewing all 40? That's exactly what Task Assessments by Charlie solves. Charlie doesn't just ask behavioral questions. Charlie gives candidates real work during the screening call: → Senior Java Engineer? Architect a microservice → DevOps Engineer? Write an automation script → QA Lead? Design scenario-based test cases The AI evaluates execution quality in real-time and delivers scored insights with detailed analytics. The transformation? Quality candidates surface in week 1. Not week 6. No more guessing based on resumes. No more wasting 40+ hours on interviews. Just a pre-qualified pipeline of candidates who've already proven they can do the work. 65% reduction in L1 screening time. Zero manual work from your team. Because credibility isn't on a resume. It's in the work itself. Curious how this works? Drop a comment or DM us. #ProductUpdate #Recruitment #TalentAcquisition #AIinHR

Hiring scam: teams asking for free work during technical rounds, then rejecting candidates right after. When a company merges your code into their product and still says no, it changes how you look at every other hiring process. This is also why take-home completion rates have dropped to around 62 percent. Many candidates don’t even finish them now. So what’s the middle ground? Candidates should decline long unpaid projects and stick to short tasks that show real skill. Recruiters should keep tests under two hours, pay for anything longer, and be upfront about how the work will be used. This is exactly why we built Fabric’s Pair Programming round. It replaces multi-day assignments with a fast, structured technical conversation. Anushka (our AI interviewer) talks through the problem, challenges decisions, and understands how someone thinks in real time. - Teams get a strong early signal. - Candidates feel respected. If your hiring still depends on long processes, the answer isn’t more steps. It’s fixing the first one. Try Fabric for free today. Demo link in the comments.

I’ve applied for many, many jobs. I rewrote my resume, tailored every application - and still… no interview invitations. At first, I thought maybe I wasn’t good enough. But then I started talking to other developers - smart, talented engineers with years of experience — and they were all saying the same thing. So no, it’s not just me. It’s the system. The current tech hiring process is… broken. You can have real project impact, strong fundamentals, solid references - and still never make it past an algorithmic filter. You spend years learning frameworks, mastering AWS, databases, clean code - and your resume gets rejected in under 10 seconds. Not because you lack skill. Because a bot didn’t find the right keyword. What’s ironic? The same companies that reject you keep posting “We’re hiring passionate engineers!” Passion isn’t the problem. Visibility is. We’ve built a hiring culture where automation decides who deserves a conversation. And that’s sad - because many of the best engineers aren’t the loudest on LinkedIn or the ones with fancy job titles. They’re the ones building quietly, solving problems, and waiting for a fair chance. So if you’re in this phase - sending applications into the void - please remember this: Your skills haven’t lost value. You are not invisible because you lack talent. You’re just stuck in a system that prioritizes filters over humans. Keep building. Keep learning. Keep showing up. One real connection can change everything.