Abodunrinwa Toki

A friend of my colleague from Google, who is based in the US, finished interviews for a SWE Product E5 role at Meta London last year. Sharing his interview experience for everyone preparing right now. Btw, if you’re preparing for system design/coding interviews, check out our mock interview tool. You can use it for free here: https://lnkd.in/gpCn7t2T 1. Phone screen – pure DSA 45 minutes, classic Meta style Q1: 3Sum   - Solved it cleanly with a clear two-pointer approach   - Dry ran on examples and called out edge cases   - Got a follow-up about avoiding duplicates, and he explained the idea, but did not fully implement the follow-up Q2: Simplify Path style problem   - Variant of canonical path simplification, current directory plus `cd`   - He struggled at the start, asked plenty of clarifying questions   - With the interviewer’s nudges, he got to the optimal stack based solution and dry ran it This was still enough to clear the screen, mainly because of clarity, communication and his willingness to ask questions instead of guessing. 2. Onsite Coding Round 1 Q1: Merge Sorted Array   - Straightforward, did in place merging from the end   - No hints, clean dry run, good variable naming and explanation Q2: Diameter of Binary Tree   - Used DFS with height calculation and a global max diameter   - Again, no hints needed, walked through tree examples and explained why the solution is O(n) 3. Onsite Coding Round 2 Q1: Sliding window average   - Given a list of integers, return list of averages for window size k   - He reached the optimal sliding window solution quickly   - No hints, careful dry run, spoke through overflow and edge cases Q2: Missing Element in Sorted Array   - Had seen it before but forgot the optimal approach   - Spent time asking questions, then realised sorted input and that binary search is the right tool   - While coding, a few bugs popped up during dry run   - Interviewer gave a hint, he fixed it, but the clock ran out 4. Product Architecture / System Design Prompt: Design Dropbox or Google Drive - Interviewer pushed on synchronization between multiple machines - He walked through trade offs between - Request response - Server sent events - WebSockets - Talked about eventual consistency and conflict resolution at a high level 5. Behavioral Round 1 and 2 - Two separate behavioral interviews, 45 minutes each - Recruiter hinted that one might be a training round for the interviewer, but did not say which - He went in with clear stories mapped to ownership, conflict, failure, and cross team work - Both rounds felt like natural conversations - They asked fewer follow ups - He could tie back every story to impact and learning His guess is both rounds landed well from a signal point of view, but he will only know for sure at the end of team match. The takeaway is simple. At E5 level, you cannot treat DSA, coding, system design, and behavioral as four sliders where you max one and ignore the others. All four matter.

600

15 Comments

Lightspeed were one of the first companies to join the Fin beta in early 2023. Interesting customer with a complex tech stack built from years of acquisitions Fin resolves 43,000+ support requests every month for them now in 100+ countries and multiple languages.

9

1 Comment

You are in a system design interview at Google for the L5 Senior Engineer role, and the interviewer leans in and asks: “Why does Spotify keep playing when I drive into a tunnel with no signal, but YouTube Music often stops or buffers? If you were designing a music streaming system, what different design choices would lead to these two behaviors?” Here is how you break it down. Btw, if you’re preparing for system design/coding interviews, check out our mock interview tool. You can use it for free here: https://lnkd.in/gpCn7t2T We’ve added new features as well:  -Company-specific interviews -In-built interview scheduler  -performance insights & trends — Both apps look like simple music players. Under the hood, they are optimized for very different priorities. [1] Spotify style – Cache first, stream second Idea: The client behaves like a smart offline player that happens to stream. The backend is built to support aggressive prefetching. What happens when you hit play - Client requests the track from a CDN or edge node. - Instead of tiny chunks, it downloads a big buffer ahead of the playhead. - In parallel, it starts pre downloading the next few tracks in the queue. - Data is written to an encrypted cache on disk, not just memory. - Playback reads from that local cache, not directly from the network socket. What this means for tunnels and bad networks - When the network drops, the player already has tens of seconds or entire tracks cached. - Because the next one or two songs are already downloaded, you can be offline for a while and never notice. - Cold start cost is a bit higher. First play might take slightly longer, but then everything feels smooth. - It burns more local storage and possibly more data, because not every prefetched song will be listened to fully. In a design answer, you can mention - Local disk cache with eviction policies (LRU per user, per device). - Background prefetch of N upcoming tracks based on queue. - Download manager that adapts how aggressively it prefetches based on network quality and user settings. - CDN tuned for larger object delivery and range requests. - Explicit offline mode that pins playlists into cache. [2] YouTube Music style – Stream first, cache is minimal Idea: Treat audio like video streams. Cost and bandwidth are optimized first. When you hit play - Player requests audio (and sometimes video) via HLS or DASH-style chunks. - Each chunk is only a few seconds long. - Client keeps a small rolling buffer in memory, not a large queue on disk. - Prefetch of future songs is limited, because video tracks are large and expensive to fetch speculatively. What this means for tunnels and bad networks - If your connection dies, the player only has a few seconds of buffered data. - As soon as those chunks are consumed, playback stalls. - Startup can feel snappy and data usage is controlled, especially for casual listeners. - Works very well on stable networks, feels fragile in spotty coverage.

1,221

38 Comments

At the weekend, I published a new write-up on squeezing even more performance out of static sites built with Eleventy and deployed on Cloudflare ✨ In this latest post, I take the idea of compression further by pre-compressing HTML to the highest Brotli level (11) at build time and serving it at the edge via a Cloudflare Worker instead of relying on on-the-fly compression. Why it matters: • Even small edge gains matter at scale – smaller HTML means faster Time to First Byte and quicker overall loads.  • Pre-compressing during the Eleventy build shifts compute away from the edge and into your deployment pipeline.  • A lightweight Cloudflare Pages Function does smart content negotiation so browsers that support Brotli get the most efficient assets. I walk through how this works, the quirks of serving pre-compressed HTML alongside normal files, and how it fits into a modern edge-centric deployment. If you care about taking static site performance that extra mile, this might be worth a look. https://lnkd.in/e5iM62Sh

20

4 Comments

This one character saved Meta 15,000 servers. Just an innocent-looking & in C++ That tiny change cut massive CPU usage, reduced memory bloat, and saved millions. When you're operating at scale, every single line of code matters. From code optimization to high-performance systems, it's the little things that make a big difference. [C++, Code optimization, High performance computing, Memory management, Backend engineering, Scalable systems, Clean code, Meta ad service, Hot path optimization, Server efficiency, System design, Developer mindset, Ampersand in C++, Programming best practices, Copy vs reference, CPU usage reduction] #CppTips #CodeOptimization #CleanCode #DevLife #TechReels #SoftwareEngineering #BackendDevelopment #HighPerformanceComputing #MetaEngineering #SystemDesign #ScalableSystems #ServerOptimization #ProgrammingReels #FullStackDev #MemoryManagement #MillionDollarBug #BigTech #EfficiencyAtScale #CodeSmart #DeveloperMindset #CodeMatters #AmpersandPower #ProgrammingHumor #LearnToCode #EngineeringLife

19

AWS Strands Typescript SDK was one of my favourite launches from re:Invent 😍. It opens up huge potential for quickly and easily writing truly full stack Typescript Agentic workflows ✨ . I've been looking how various emerging open standards and open source tooling makes writing agentic applications easier than ever... 💪 In this blog post I've written how you can use the open protocol, AG-UI, alongside the open source Strands Agent framework and the open source Copilot Kit to write a truly full stack Typescript Agent and UI. The more I've been using the Typescript Strands SDK the more I'm convinced this will be one of the biggest impacting long term launches from re:Invent. The full Github repo is in the comments. #aws #strandsagents #bedrock #genai https://lnkd.in/e7uKc-bk

21

3 Comments

If you're a Senior Engineer looking to move up, the next role will likely be a Staff Engineer. And in this guide, Shruti shares tips from her own experience of getting promoted to Staff Engineer at PayPal and Slack. She talks about what Staff Engineers do (and how it's different from Seniors), why you might not be getting promoted, and how to take that next step. https://lnkd.in/g58dnFEG

103

3 Comments

Today we launched Model Context Protocol (MCP) support in Amazon Q Developer CLI, allowing developers to connect external data sources for enhanced context-aware responses and improved code generation. For more details please check the blog. https://lnkd.in/gm8qn84E

70

2 Comments

Beyang Liu has a great write-up over on X for why Agentic AI applications are fundamentally different than RAG... and why that means model selectors no longer make sense. Give it a read! https://lnkd.in/g9e2gGCN

13

Platform engineering is the future 🚀 Developers are at the core of everything we do at PagerDuty 💚 Our latest collaboration with Spotify for Backstage brings PagerDuty’s industry-leading Incident Management directly into Backstage, the open platform reimagining developer experience. With the new PagerDuty Plugin for Backstage, devs can see incidents, on-call context, and service ownership right where they already work : Less dilution ; More focus. Let’s go 🎉 Read more 👇 https://lnkd.in/gnsq8Nr9

28

2 Comments

Staying Close to the Code in an AI Era Midnight. EKS upgrade window in progress. 💡While monitoring cluster health and rollout stability, I used the window to clean up some legacy iOS code from our React Native repository after our migration to fully native iOS. 🤯I cloned the repo. 368 MB. Our production app binary is ~5 MB. ✅ I put Claude removed legacy iOS-specific modules from the RN repo and validated Android build integrity. A few days work done in 30 odd minutes. ✅ In parallel, with the help of one of our senior engineers, we found the root cause of 368 MB — accumulated Git object history. Years of binaries and artifacts had quietly expanded the repository size. The working tree was lean. The .git directory wasn’t. A few reflections: 💡 Small metrics often reveal systemic signals. 368 MB repo size for a 5 MB artifact isn’t a problem — it’s a prompt. 💡 AI accelerates execution. Clarity still comes from engineers thinking together. 💡Leadership doesn’t mean stepping away from the code. It means staying close enough to notice what quietly compounds. 💡Speed is increasing across our industry. Discipline has to scale with it. And the real leverage? 💡Combining AI acceleration with first-principles thinking. #AI #CodeHygiene #FirstPrincipleThinking #Engineering

35

🚀 Backend Observability in 2025: Distributed Tracing with OpenTelemetry In 2025, building reliable, scalable backend systems requires more than just good code, it demands deep observability. Traditional logs and metrics aren’t enough for today’s distributed systems. I’ve written a new Medium article exploring: ✅ Why distributed tracing is essential in modern backends ✅ How OpenTelemetry is becoming the industry standard ✅ A practical guide to implementing tracing in your services If you’re working with micro-services, server-less architectures, or edge systems or just want to level up your backend engineering, this article is for you! 👉 Check it out here: [https://lnkd.in/g_Gmyh5W] I’d love to hear how you’re using OpenTelemetry or other observability tools in your projects. Let’s discuss in the comments! #BackendEngineering #Observability #OpenTelemetry #DistributedTracing #Microservices #Backend #SystemDesign #DevOps

23

An agent has no consistent definition. You can think of an agent as an autonomous or semi-autonomous system that can take actions on behalf of the user in a given environment, state, and make decisions or take actions that can accomplish certain tasks along a given time-frame, either by calling tools, dynamically selecting the next action to take or use a deterministic "workflow" (systems where LLMs and tools are orchestrated through predefined code paths). Using an Agent framework that gives you the ability to built systems in both, a reliable and a dynamic way can accelerate your agent development journey using agent abstractions. This means building Agentic systems that can reliably call tools, store memory (episodic/semantic/procedural), have comprehensive logging and observability, human in the loop workflows and the ability to build various multi-agent patterns flexibly based on your use case. A successful Agentic system in production is usually a combination of both, dynamic and predictable/reliable multi-agent systems. Strands Agents SDK gives you exactly these capabilities by treating each “agent” as a combination of a foundation model plus a suite of tools. You define a prompt and register your tools (decorated functions) in code, then Strands handles reasoning→planning→tool-execution cycles, local testing, and cloud deployment (ECS, Fargate, Lambda, EC2), along with support for all other agent abstractions provided above. Excited to share a hedge-fund analyst multi agent system: This uses the newest Anthropic's Claude 4 Sonnet/Opus that powers the Lead Analyst Agent, routing incoming queries to specialized sub-agents for fundamental, technical, and market analyses. Each specialist is wrapped as a callable tool (using the “agents-as-tools” multi-agent pattern), so the orchestrator never has to implement domain logic itself and can handoff the task to an agent as a tool. For sensitive operations (insider lookups), we utilize a HITL approval step that halts execution until a human grants consent. We also use meta-tooling that enables the Lead Analyst to generate, load, and invoke new custom tools at runtime—whether it’s a portfolio beta calculator or a pricer—without redeployment. Strands also embeds observability (Langfuse) and OpenTelemetry tracing so you can trace reasoning events, tool invocations, errors, and end-to-end workflows in real time. View more information on the code implementation here: https://lnkd.in/gJmwVyGi Code implementation: https://lnkd.in/gzTtJvJq Thanks to 🏄‍♂️ Cagatay Cali for being a reviewer/collaborator on this! Feel free to try it out and reach out with any questions/ideas. #aws #agenticAI #strands #agents #generativeAI

60

6 Comments