Engineering Competency-Based Interviewing

I've interviewed probably close to a thousand people for ML/AI engineering roles since 2016. This is how I find out who can actually ship. Rule #1 Do not ask questions which have a correct answer. They either know a fact or they don't. Who cares. Rule #2 Make them the expert, then get relentlessly curious. I ask them to describe a past project and just keep pulling at a thread. How did that *actually* work? Keep asking harder questions and get to the edge of their knowledge. Don't be afraid to ask questions you wouldn't know how to answer yourself (the truly great candidates will change how you think) Then I talk them through a hypothetical. One of my favourites: You have a dataset of ten thousand 30-second video clips. The task is to predict the final 10 seconds given the first 20. Which is harder to predict: the video or the audio? Some people talk about data dimensionality. Some talk about tensor operations. And some people ask: "well, what's happening in the videos?" You have 24 hours to decide if this project is feasible. What do you do? Some say "go read the literature." Some go look on huggingface for a pretrained model. And some would spend 20 hours watching the videos. We've decided to take on the project. What's the dumbest possible baseline you can build? Can they come up with something OTHER than gradient descent? It's a high-signal, no-BS interview that's helped some really stellar people stand out from a pile of CVs. And it's chatgpt-proof, too.

In the last eight years, I have interviewed 500+ Software Engineers for various roles. Here are the most actionable tips I can give you on how to do better during your behavioral round. 1/ Set the Stage Clearly - Describe the Situation or Task that needed solving. Focus on the challenge. - Example: "The API response times were too slow, affecting user experience, and I was tasked with optimizing it within a sprint." - Keep it short. If the interviewer wants more details, they’ll ask. 2/ Focus on Key Actions - Highlight 3 core actions you took to solve the problem. - Example: "I profiled the API calls, implemented caching for frequent queries, and reduced payload size by 30%." - Stick to impactful actions. Each action should take under 2 minutes to explain. 3/ Use “I” to Show Ownership - Make it clear what you did to demonstrate leadership and initiative. - Example: "I spearheaded the migration from monolithic architecture to microservices, improving scalability by 40%." - Avoid saying "we" too much. The interviewer needs to know if you led the effort or just contributed. 4/ Stick to Facts, Avoid Emotions - Keep your answers factual, even when discussing challenges. - Example: Instead of "I was frustrated with a teammate’s slow progress," say, "I scheduled a pair programming session to help them meet the deadline." 5/ Understand the Purpose of the Question - Think about what the interviewer is trying to assess—teamwork, problem-solving, leadership, or technical expertise. - Example: If asked about handling conflict, they want to see how you navigate disagreements productively. Frame your response accordingly. 6/ Use Data to Back Your Results - Quantify your impact wherever possible. - Example: "After optimizing the query logic, I reduced database read times by 40%, cutting down page load times by 2 seconds." - Data shows real impact and demonstrates the value you bring. 7/ Keep It Interactive - Make your responses concise to encourage follow-up questions from the interviewer. - Example: "I automated the deployment pipeline, cutting release times from 2 hours to 15 minutes. If you'd like, I can explain the challenges I faced setting up the CI/CD tools." 8/ Maintain good eye contact -Eye contact showcases confidence -In the era of online interviewing, it’s even more critical to showcase your focus via eye contact. And one thing you should never do in the behavioral interview is makeup details. It’s visible how shallow a story is if someone grills you on the details. I hope these tips will help you achieve great results. – P.S: Follow me for more insights on Software engineering.

Great Engineers Don’t Memorize, They Solve. I once took a group of technical hiring managers out to lunch, and I couldn't help it, I had to ask... “What’s with the brutal technical interview questions? Then a long pause and I asked...Could you answer any of these if you were on the other side of the table?” They looked around at each other… smiled… and then burst out laughing. “No,” they said. Yes, you read that right. So why are we still grilling candidates like it’s a certification exam? Interviewing is nerve-wracking enough. If you want to hire smart, capable engineers, ditch the trivia and focus on how they think, how they troubleshoot, and how they work in real environments. Instead of asking: “What port does X run on?” “What’s the difference between symmetric and asymmetric encryption?” Ask: “How would you respond if a system alert flagged a critical issue?” “Tell me about a time you made something more secure or more reliable.” “How do you approach solving something you’ve never seen before?” You’re not hiring a walking manual. You’re hiring a problem solver. Interview for that. If you’re already doing this, great!! We need more of it. #techhiring #engineering #interviewstrategy #recruiting #TheInsideGame

At Weave, we interviewed 134 people to find our 2 founding engineers (shoutout to Drew and Steven!!). One key thing we looked for was AI competence. Here's a 3-step breakdown on how to screen for it: First of all, why does this even matter? Because I see lots of companies hiring the wrong people because they're testing 2019 skills. The best engineers in 2025 aren't the ones who memorize algorithms. They're the ones who combine raw intelligence and quick learning with AI mastery. AI has changed how we code, so the process to assess someone's ability to code needs to change as well. 𝗦𝘁𝗲𝗽 𝟭: 𝗧𝗵𝗲 𝗧𝗮𝗸𝗲-𝗛𝗼𝗺𝗲 𝗔𝘀𝘀𝗶𝗴𝗻𝗺𝗲𝗻𝘁 Give them a realistic coding challenge with explicit permission to use any tools they want. We use NextByte (YC W25) at Weave which makes this really easy. But if you're not using a tool like NextByte, you need to schedule a 30-minute follow-up where they walk through their solution. Ask them to explain their approach, then drill deeper on reasoning. Ask "Why did you do that?" or "How does that work?" until they can no longer answer. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸𝘀: Everyone uses AI on take-homes anyway. By allowing it explicitly, you can focus on what actually matters - do they understand the code they're submitting? Can they explain design decisions and trade-offs? How deep is their knowledge of the technologies they work with? The drilling reveals whether they have deep technical understanding or just copy-pasted AI output. Plus, this makes the take home more accurately reflect how their actual work will look. 𝗦𝘁𝗲𝗽 𝟮: 𝗦𝗽𝗹𝗶𝘁 𝗬𝗼𝘂𝗿 𝗟𝗶𝘃𝗲 𝗜𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄 𝗜𝗻𝘁𝗼 𝗧𝘄𝗼 𝗣𝗵𝗮𝘀𝗲𝘀 Phase 1: Problem-solving without AI (40 minutes) - test reasoning on system design. Phase 2: Implementation with AI (20 minutes) - watch how they work with AI tools. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸𝘀: You need both skills. Phase 1 reveals their ability to think through complex problems and architect solutions. Phase 2 shows whether they can efficiently leverage modern tools. Engineers who excel at only one phase will struggle in real work environments where both matter. 𝗦𝘁𝗲𝗽 𝟯: 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗲 𝗧𝗵𝗲𝗶𝗿 𝗔𝗜 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝘆 Ask them to walk through their typical workflow and explain when they choose AI vs. manual coding. Ask about specific projects they have built. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸𝘀: The best engineers have developed intentional approaches to AI usage. They know when AI helps and when it hurts. They can explain their prompting strategies and understand AI limitations. If you want to hire a team that can thrive in 2025 and beyond, LeetCode doesn't cut it anymore!

Wanted to share the game-changing document from Andreessen Horowitz that transformed how we hire! When we raised our seed round with a16z, they shared a hiring tool that changed our recruiting approach: the MOC Document. Before you ask, yes I will happily share the document! But first, let's walk through how it works. 💡 Context Matters Everyone knows that hiring the right people is crucial to growing a business. However, who doesn’t struggle with getting it right? I used to rely on job descriptions and scorecards. They were good, but not great. With many business critical roles to fill and countless candidates to consider, we needed a better way to filter prospects along metrics that actually predicted future success. 🔑 Enter the MOC Document The MOC (Mission, Outcomes, Competencies) document is a detailed blueprint for every hire. It's not just a job description — it's a strategic guide that aligns everyone on what each role needs to achieve. Here's a breakdown: A) Mission: Define the role's purpose B) Outcomes: Specify measurable goals C) Competencies: List required skills and behaviors with sample interview questions to gauge fit. Let's walk through an example. 📊 In-Depth Example from our Head of Engineering Role: Mission: Lead and scale the engineering team to become a 99th percentile high-performance team that values excellence, teamwork, and solving hard problems. Outcome (1 out of 3) - Team Development: A) Goal: Hire talented engineers and engineering managers to enable the product's success, ensuring technology talent is not a bottleneck to our success. B) Metrics: 1) Increase our engineering org from 25 to 50+ in the next 12-18 months. 2) Maintain attrition below 10%. 3) Foster a thriving culture with a strong engagement score (80+) in pulse surveys. Competency (1 out of 8) - High-Impact Organizational Leader: A) Description: Actively shaped their past engineering organizations to have a strong culture of high-output teams, created the right organizational structure to accelerate development, and showcased high agency in performance management. B) Sample Question (1 out of 5): "Was there ever a time where a team felt disengaged? How come that happened? How did you diagnose the problem? How did you turn it around?" C) Strong Yes Criteria: Has built highly-engaged teams that thrive in the culture that the leader was shaping. Has great examples of turning around less-engaged teams. 🌟 Why the MOC Document Works: Creating this document takes multiple hours for each position. However, it’s worth it. It helps us fully align on what the role needs to achieve and keeps us focused on what to look for at each step of the hiring process. The level of clarity provided by the MOC sets the foundation to attract stronger candidates, shortens the hiring process, and better integrates new leaders into the organization. Want a sample MOC document? Drop your email below and I’ll send it your way! 📩

A few years ago, I realized most recruiters were winging it when talking to technical candidates. They’d ask generic behavioral questions or skip the technical part entirely — and the best candidates could tell instantly. I’ve been there too. Early in my career, I remember fumbling through my first technical interviews, trying to sound confident while barely keeping up. It wasn’t until I started mapping the interview to how engineers actually work that everything clicked. After training 3,000+ teams on how to engage and qualify technical talent, I built a simple 5-step framework that works every time. A 5-step framework borrowed from the UX toolkit that gives recruiters confidence and consistency — even when the role feels intimidatingly technical. No gimmicks. No AI content slop. Just a framework that maps directly to how technical professionals actually work. Here’s the system 👇 Every technical candidate follows a predictable 5 step workflow — Software Development Life Cycle (SDLC). When you understand this workflow, you have a roadmap for your interviews. Think of it like a journey map for recruiters: a visual of how the candidate builds solutions, step by step. Here’s the SDLC — and your built-in interview outline 👇 1️⃣ Research – How do they uncover what’s needed for a feature or product? 2️⃣ Design – How do they translate insights into a potential solution? 3️⃣ Build – How do they actually create or code it? 4️⃣ Test – How do they evaluate success or identify issues? 5️⃣ Deploy – How do they push their work live and support it in production? This one insight changed everything for my clients. They stopped guessing. They started connecting — and qualifying with clarity. Because when you interview with the SDLC in mind, you’re speaking the language of engineers. Want me to break down each of these 5 stages with example questions you can use? Drop a 🔥 in the comments and I’ll share the templates.