Konrad Cheng

How long does it really take to reach Senior as a FAANG Software Engineer? Looking at US FAANG SWE salary submissions to Levels.fyi from the past two years reveals some interesting patterns in the path to Senior Engineer and how it compares industry-wide. Across all companies in our database (not just FAANG), the average SWE reaches Senior at around 8.6 years of experience. We roughly calculated this by looking at the total “Years of Experience” of our US SWE submissions and subtracting the “Years at Level” value from it to approximate when someone was hired or promoted into Senior. So how does FAANG compare? - Amazon: Slower to promote. Even with 10–12 years in the industry, only ~50% are Senior+. A big reason is Amazon’s hiring approach, where many mid-career engineers come in at L5 (SDE II). That creates a large pool of experienced engineers who may sit at L5 for years, making the climb to Senior look slower. - Google & Apple: Faster paths, with most engineers Senior by 7–9 YoE. - Netflix: Almost everyone with 10+ YoE is already Senior, reflecting their flatter org structure and strong bias toward hiring directly into senior roles. Very few engineers with significant experience show up below Senior in our data. - Meta: Falls somewhere in between, with a wider spread of outcomes. Taken together, FAANG engineers tend to reach Senior a bit faster than the **industry-wide 8.6-year average**. But the differences across companies are stark. Some ladders are structured to get you there in under a decade, while others keep a broad mid-level population for much longer. Check out more here: https://lnkd.in/gNAuv-gC

242

4 Comments

Imagine you're sitting in an interview for a system design interview with a Meta Staff for an E5 role, and you're asked this question: "Design Leetcode." I'm a Principal Engineer with over 12 years of experience. Here's how I would break it down if I were the one being interviewed:  [1] Start with requirements, not Redis ➤ Clarify functional flows in 60 seconds   -- View list of problems   -- View a single problem with description, examples and code editor in browser   -- Submit code and get verdict for each submission   -- Run timed competitions with fixed start time, duration and problem set   -- Show a live leaderboard for each competition ➤ Lock in non functional expectations   -- Peak load is around 100k users in a contest window   -- Leaderboard should feel near real time, updates every 2 to 5 seconds are fine   -- Availability over strict consistency, a slightly stale leaderboard is acceptable   -- Low latency reads for problems and leaderboard, verdict latency within a few seconds ➤ Make security and isolation a first class requirement   -- We execute untrusted user code on our infrastructure   -- User code must never see our main database, file system or internal network   -- We need sandboxing, timeouts per run and strict CPU or memory limits ➤ Get concrete scale numbers from the interviewer   -- Daily active users and contests per day   -- Problems count, average submissions per user per contest   -- These numbers will justify why we add queues, caches or container clusters ➤ Write this requirements box before drawing any components   -- Functional list   -- Non functional list   -- Scale assumptions   -- Then say clearly, “Everything I design next will be to satisfy these constraints.” The next parts: [2] Core entities and APIs, [3] High level architecture, [4] Code execution isolation, [5] Live leaderboard with Redis sorted sets, [6] Scaling contests with queues and workers Are covered on the blog I’ve put in the comments, take a look there. — P.S: Feel free to reach out to me if you're preparing for a switch, want to chat about interview preparation, or how to move to the next level in your career: https://lnkd.in/guttEuU7 For Mock interviews: https://lnkd.in/gKWbHmke

930

51 Comments

𝗪𝗲𝗲𝗸 #11: 𝗜𝗳 𝗜 𝗴𝗼𝘁 $1 𝗲𝘃𝗲𝗿𝘆 𝘁𝗶𝗺𝗲 𝗜 𝗱𝗼𝘂𝗯𝘁𝗲𝗱 𝗺𝘆𝘀𝗲𝗹𝗳… 𝗜 𝘄𝗼𝘂𝗹𝗱’𝘃𝗲 𝗿𝗲𝘁𝗶𝗿𝗲𝗱 𝗮𝘀 𝗮 𝗺𝗶𝗹𝗹𝗶𝗼𝗻𝗮𝗶𝗿𝗲 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 🥲 “Am I really that good compared to what people say?” “If I had to restart my career tomorrow because of some visa issue, could I still land a job?” “If someone asked me what I’m good at as a Software Engineer… what would I even answer?” Self-doubt. Insecure. Imposter syndrome. Whatever name we put on it, it all feels the same: not fully believing in yourself. And for me, it shows up in my career as a Software Engineer. Strange thing, though: I don’t feel this way when I do Leetcode or OAs. Why? 1. I know the rules. There’s a syllabus: common data structures and algorithms. 2. I practiced. 300+ questions. Each timed. Each with notes like I’m in a mock interview. 3. I stacked small wins. 300 solved = 300 wins. Even the unsolved ones count as a win when I at least had a direction. But at work? ✔ I practice daily — coding, testing, debugging, monitoring, learning. ✔ I stack small wins — I’ve kept 200+ daily journal pages since joining Microsoft last July. Still… ❌ I don’t know the rules. Or to be more correct: I never really tried to look for the “syllabus.” I just went with the flow, picked things up as they came, and hoped it was enough. But that uncertainty? That’s what feeds my doubt. That’s the missing piece. So maybe the framework I will need to follow moving forward is simple: 1️⃣ Learn the rules — understand how the industry game works and what the core principles are. 2️⃣ Practice, practice, practice. 3️⃣ Keep stacking the small wins. One more thing — I’m lucky to have amazing colleagues and managers. Even on days I doubt myself, I know they believe in me. And that really helps. Maybe you have people like that in your corner, too. Y̲o̲u̲r̲ ̲t̲u̲r̲n̲: What have you done lately to build confidence? Drop it in comments — your win or struggle might help someone who feels the same. #𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴𝗢𝘂𝘁𝗟𝗼𝘂𝗱𝗪𝗶𝘁𝗵𝗧𝗿𝗶 #𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝗝𝗼𝘂𝗿𝗻𝗲𝘆

6