Emilian Miron

From IC engineer to C-Level exec, the most effective people I've worked with have one thing in common: they break down and decouple decisions. Optionality, trade-offs, last responsible moment, one-way and two-way door decisions, slicing, MVPs, proof of concepts etc. are all techniques they use for breaking decisions down and getting them made. The best engineers did this when working on code. Rather than trying to decide on all the functionality at once they'd be able to say "I can break this story down into smaller pieces of value that we can get released to users fast" or "we don't need to deliver this piece of functionality now, we can wait until we see how user's respond". The best architects, product managers and business leaders do the same thing. It's a real skill to decouple decisions. It's a real skill to be able to defer parts of a decision to next or later (or even never) to enable faster movement and lower risk. I have always been impressed when I witness it. Often it's obvious with hindsight and sometimes its completely out-of-the-box and you would have never have realised.

60

3 Comments

Everything else being equal, finding bugs statically is better than finding bugs at runtime. Everything else is rarely equal. In the comments to my last post, someone asked about the benefits / adoption of formal verification compared to testing. I have some-but-not-a-ton of experience with formal verification, but I have a lot of experience with static analysis tools. For those of you who don't know, formal verification involves making a mathematical specification for what your code is supposed to do, and then running a tool that can prove it meets that specification. Most people reading this post will now know why most developers don't do it. (It's worth a reminder - tests *are* a specification for your code. They're just one that's written in the language in which you wrote the test instead of being written in math.) It's a truism that it is better to find a bug at compile time than at test time, at test time than at code review time, and at code review time than at deployment time. Each additional step bakes in problematic decisions, giving them the chance to spread to other code, and, eventually, to users. Also, every second that passes between when you write the code and when you notice the problem is a second your brain might use to forget what the code does. I wrote a post a couple of months ago about research done at Google to identify what the ideal static analysis tool looks like. In short: you don't have to do any work to make it run, it never gives you false positives, it points out real problems, and the tool tells you exactly what you need to do to fix the problem. Tools like this are great, and are why Google leans heavily on govet, clang-tidy, and Error Prone (errorprone.info). The more typing that someone has to do to make a static analysis tool work, though, the more that they are likely to write a test, instead (possibly adding a dynamic analysis tool). Formal verification tools are at the far end of this - they tend to make you write specifications for your code, which they then prove the code obeys. Everyone who has worked on a codebase in the broader tech industry knows that most specifications last (at most) as long as it takes to write the code, at which point you realize you got them wrong and change them. This makes formal verification a lot of work for very little reward. The only people who write formal specifications for their code are the ones who are writing code that will kill you if they get it wrong - code to keep airplanes flying or to keep nuclear reactors from meltdown. Most Java development is way back at the other end, where we still beg and plead with people to add nullness annotations (it will help when the jspecify.dev project finishes). This is why sneaking static analysis features into programming languages is so important (I hear you, Kotlin fans). At some point, people's interest in typing annotations drops off, and for everything beyond that, we have tests.

35

10 Comments

sharpening writing skill is one of killer skill!

2

I am excited to share the latest blog, co-authored by myself, Junhwi Kim and Vyas Sekar, detailing how we managed to successfully localized and improved LLM by combining Fine-Tuning and RAG. https://lnkd.in/gKDkgGPq

69

3 Comments

Google SRE NYC Tech Talk - May 22nd, 6:30pm Google SRE NYC proudly announces our second Tech Talk event in 2024. The in-person only event will take place on Wednesday, 22nd of May 2024 at 6 PM at our Chelsea Markets office in NYC. The doors will open at 5:30 pm. Spaces are limited, RSVP via our Meetup event page. https://lnkd.in/ef5fqdrE Agenda: Salvatore Furino - Customer Reliability Engineer (CRE) at Bloomberg “The Hammer Changes the Hand” This talk will briefly explore how to view internal tooling through the lens of product management in not just developing and shipping features, but how those features empower teams to change their understanding of their social-technical systems. Sal is a Customer Reliability Engineer. During his career he’s worked as a TPM, SRE, Developer, Sys Admin, and IT support. While not working he enjoys cooking, gaming, traveling, skiing, and golfing. Sal lives in Queens and has a BS in Applied Mathematics from Marist College. Thiara Ortiz - Staff CDN Reliability Engineer at Netflix  “How we measure Quality of Experience to ensure our members get a world class experience they have come to expect from Netflix ” Any time a Netflix member sits down, reclines in their chair and turns on their TV to Netflix, there's a moment of truth. It's an opportunity to deliver a spectacular service with amazing quality of experience. This talk will go over how we measure the quality of experience for our members and how we work to develop new metrics when we have additional offerings like live streaming and cloud gaming. Thiara has worked at some of the largest internet companies in the world, Meta and Netflix. During her time at Meta. Since Meta, Thiara has been working at Netflix as a Staff CDN Reliability engineer. Her focus is primarily on resilience and quality of experience for members streaming from Open Connect. When incidents occur and Netflix's systems do not behave as expected, Thiara can be found working and engaging the necessary teams to remediate these issues. Mike Scherbakov - Staff Site Reliability Engineer at Google “LLM for SRE / Using LLM in SRE space” LLMs open up an opportunity to automate and scale many operational processes, which couldn't be otherwise solved by conventional methods. Examples include simple summarization of issues and incidents, assisting production on-callers, managing incidents, clustering (creating taxonomy) of issues, scaling SRE via assisted review of development design documents. Therefore LLMs provide a new and unique opportunity to transform the work we do as SREs. Mike works in YouTube Ads SRE as well as co-leading LLM4SRE, LLM center of competency in SRE at Google Our Tech Talks series are focused on professional development and networking: no recruiters, sales or press are allowed.

16

3 Comments

I'm excited to announce that I'll be presenting at the Codeborging: AI-Powered Programming Meetup on June 12 on "Harnessing OpenAI Assistants for Dynamic Content Generation." In this session, I'll share insights from my journey of using OpenAI Assistants to streamline content creation and enhance productivity. Key highlights will include: - Knowledge Bot: A context-driven OpenAI Assistant that personalizes content generation using advanced GPT-4o models. - Amazon Treasures Chat: Enhancing the e-commerce experience with AI-driven conversational interfaces. We will dive into practical code examples, demonstrating how to set up these assistants, craft effective prompts, and integrate them into your projects. Key Takeaways: - How to prepare data and configure OpenAI Assistants - Practical Tips for Generating Natural Language Search Results - Real-world applications to boost daily productivity, like generating blog posts, handling communications, etc. Join us and unlock the potential of AI in your projects. Don't miss this opportunity to elevate your content strategy and productivity with AI! #AI #ContentCreation #OpenAI #TechTalk #MachineLearning #Productivity #SoftwareDevelopment Artium LeanDog

12

1 Comment

May 22nd 2024 Google SRE NYC Tech Talk recap Google SRE NYC hosted our second Tech Talks meetup for 2024 last week. https://lnkd.in/ef5fqdrE It was great to meet new community members for the first time and welcome back many familiar faces in person! Our Meetup community ticked over 1K members and continues to grow - a testament to how active and engaged the NYC SRE / DevOps community is, and a great feedback datapoint that motivates us to keep going with our events series 🙂! Huge Thank You to our wonderful speakers for their hard work developing and presenting their very interesting and relevant tech talks! If you are interested in speaking at our future events, or know someone who might be interested, pls fill out this speaker registration form (https://lnkd.in/eEnYX7vm). If you have something in mind, but not quite ready yet, don’t worry! We still have at least two more events this year, the next event is likely to be in Sep and one more in Dec. Our content reviewers are ready to guide you through the process and offer helpful feedback. If you couldn’t make it to the May 22nd event, here’s a quick recap of what you’ve missed: Our audience enjoyed talks from: Salvatore Furino - Customer Reliability Engineer (CRE) at Bloomberg “The Hammer Changes the Hand” The tools we use change us and enable us to think and interact with the world differently. A brief exploration of how to view internal tooling through the lens of product management in not just developing and shipping features, but how those features empower teams to change their understanding of their social-technical systems. Thiara Ortiz - Staff CDN Reliability Engineer at Netflix “How we measure Quality of Experience to ensure our members get a world class experience they have come to expect from Netflix ” How we measure the quality of experience for our members and how we work to develop new metrics when we have additional offerings like live streaming and cloud gaming. Mike Scherbakov - Staff Site Reliability Engineer at Google "LLM for SRE / Using LLM in SRE space" LLMs open up an opportunity to automate and scale many operational processes, which couldn't be otherwise solved by conventional methods. E.g. simple summarization of issues and incidents, assisting production on-callers, managing incidents, clustering (creating taxonomy) of issues, scaling SRE via assisted review of development design documents. Therefore LLMs provide a new and unique opportunity to transform the work we do as SREs. As always, after the Q&As, the attendees mingled with the presenters and each other over some refreshments and gourmet snacks! If you attended one or more of our events, please share some feedback. If you have any questions, suggestions or comments pls message us via Meetup chat / comments or DM me :). Vlad

16

1 Comment

In the era of LLMs, "sexy" might not be the term, but bag-of-words solutions are still kicking! When the position of a word matters, consider using Positional Vectorizer over TFIDF. Check it out: https://lnkd.in/dDRYFrAX

14

1 Comment

One of the incredible topics I've been able to learn about, and teach, at Amazon has been implicit communication - the meaning that HOW you write and say things imparts to WHAT you write and say. Along the way, I've found quite a lot of information in the form of research studies, scholarly articles, books, factoids, tips, tricks, and so forth - like this article about how being facially expressive has been shown to make people more likeable and socially successful. However, the most valuable lesson I've learned isn't any of those tips, tricks, or facts. It's the importance of being empathetic and authentic: to understand my own emotions and adjust my reactions in response to others, and to instill in myself a genuine desire to collaborate that aligns my implicit and explicit communication. So much of implicit communication is hard-wired into us as humans and happens subconsciously in ways we can't control - but we can control our own emotional state and awareness of others' needs. What emotional state do you want to have going into any conversation or situation? How can you evoke that emotional state in yourself? How can you identify when you may have misinterpreted the communication of others? What can you do to address and mitigate the impact of that misinterpretation? #amazon #leadership #emotionalintelligence #communication #collaboration

18

2 Comments

An open letter from Meta employees to Mark Zuckerberg and Meta's leadership. Very well articulated, I couldn't have done it myself better! I left Meta end of 2023 because of many things similar to what's mentioned in the letter, so I can definitely relate.

87

I'm excited to share some great news. The startup I joined less than 2 years ago, Anonym, has been acquired by Mozilla! I couldn’t be more excited to join Mozilla and work with the amazing teams there. Like Anonym, Mozilla believes that we need a free and open internet supported by digital advertising, and that digital advertising doesn’t need to compromise people’s privacy. You can learn more about the acquisition by reading Mozilla's blog post (https://lnkd.in/eVkbmzkb) and the AdExchanger article (https://lnkd.in/euckaKPR).

40

5 Comments

I started writing regularly about AI. If you're an innovators or product developer eager to apply the latest in AI, whether you're in a startup or a big company, consider checking https://lnkd.in/gGdW8PYe I won't teach you ML, but I share real-world insights from launching - and sometimes failing at - numerous ML features on a large scale.

55

2 Comments

After a year of work my book Building A Startup - A Primer For The Individual Contributor is finally available! https://lnkd.in/erThxvpP If you’re working in or around startups as an IC or technical leader this book is for you. It’s based on 20 years of startup work and will help set you up for success to do more with less, move faster and more correct, and spend more time building features and product people love and less time banging your head at roadblocks and bottlenecks.

115

12 Comments

TL;DR Impacted by the Tesla layoffs. A wild ride worth reflecting on and taking some time to recharge the batteries. It all started innocently enough with a LI message from a Tesla recruiter in the early months of 2020, and little did I know the entire world would be turned upside down for the better part of 2 years. What was supposed to be an onsite interview suddenly became a virtual interview, and ended up as a WFH onboarding. Tasked with building a scalable Learning platform from scratch for our almost 160k+ strong global workforce of employees and partners as we migrated systems with an early 2021 delivery deadline seemed an almost insurmountable ask at the time, very much complicated by being a new hire and trying to get things done in a virtual work environment the entire company was figuring out. Against this backdrop, I got to witness first-hand the undeniable dedication, efficiency, and passion of all of my colleagues at Tesla. There was no hesitation when we had to spend late nights until 3am during Thanksgiving week nor the countless weekends collaborating to ensure we got the job done. And got the job done we did. Fast forward 4 years, and together we’ve built an enterprise Training platform that has scaled to support Tesla’s almost 300k global workforce of employees and partners through eLearning, In-Person instruction, and on-the-job Assessments. We had the unique opportunity of supporting all parts of our business globally from HR to Engineering to EHS&S to Energy to Manufacturing to SS&D and to Supply Chain. While it’s certainly a disappointment for myself as well as for my many dedicated and passionate colleagues who have been impacted, I take immense pride in knowing that we all contributed to a greater cause and mission than ourselves: Accelerating the world’s transition to sustainable energy. And even though we’ll no longer carry the torch for now, I know that through our many contributions that we’ve left Tesla in a better place than when we joined, and Tesla is absolutely primed for the next growth phase. We’ll all be cheering from the sidelines! I would be remiss if I did not mention some of the talented colleagues who were instrumental to my Tesla journey: Malia S. Eben Carey Dave Parrent Shelby Callahan Avinash Matha Janelle Flores Lixin Wang Prateek Khanna Pankaj Saini Ghizlane Bounoua Aarthi Palaniswamy Renee Hu Lizette B., PsyD Molly Caddam David Dodge, PMP Dillon Lee Matt Primeau Jack Sylvester Kevin McMurphy Alfredo Aguirre Stephanie Hernandez Divya H Manasvi Komma Mike Nobles Carly Bermensolo Rafae Shaik Xiechuan Liu #Tesla #Teslalayoffs #TSLA

476

37 Comments

Very excited to share that I've completed an enriching Natural Language Processing(NLP) course at UC Berkeley this quarter(Part of my Masters degree)! Delving into the intricacies of language comprehension and analysis has been truly enlightening. As we explored concepts like information extraction, sentiment analysis, and machine translation, I couldn't help but reflect on the significance of attention in NLP. Indeed, "Attention is all you need"(Transformer based model) proved to be a guiding principle throughout our studies. Grateful for the opportunity to expand my knowledge in this field! #NLP #UC Berkeley School of Information I highly recommend diving into the research paper linked here (https://lnkd.in/ggA_SrUb), though it may take a few reads to fully grasp its depths, along with Jay Almmar's insightful blog post (https://lnkd.in/gbDJ3Ddj). Excited to delve into "Generative AI" as my focus for the upcoming quarter!

61

5 Comments

🔍 Demystifying Distributed Systems: Raft's Secret Sauce for Reliability As an engineering manager, I've seen firsthand how critical consistency is in distributed systems. Today, let's unpack the Raft consensus algorithm's approach to leader election and consistency. Trust me, it's simpler (and cooler) than you might think! 🎭 The Players: Leaders and Followers Imagine a high-stakes game of "follow the leader" where servers are the players. In Raft: One server is the leader The rest are followers But what happens when the leader stumbles? 🏁 Leader Election: A Race Against Time The leader sends regular heartbeats to all followers If followers don't hear a heartbeat, they call an election Followers nominate themselves and request votes The first to get a majority vote wins and becomes the new leader But here's the kicker: Not just any follower can become leader. They need to prove they're up-to-date enough to lead. It's like ensuring the new team lead knows about all the recent project decisions! 🔒 Consistency: Keeping Everyone on the Same Page Once we have a leader, how do we ensure all servers agree on the data? All changes go through the leader The leader replicates changes to followers Once a majority confirm, the change is committed The leader then tells everyone it's official Think of it as a team decision: once most agree, it's locked in, and everyone gets updated. 💡 Why This Matters: In the world of microservices and distributed databases, Raft's approach ensures: There's always a clear decision-maker Data remains consistent even if some servers fail The system can recover quickly from failures Real-world impact? Imagine an e-commerce platform that never loses your cart items, or a banking system that always knows your correct balance, even during server updates! 💾 Real-World Database Examples Different distributed databases handle consistency in various ways: 1. etcd & Consul: Use Raft for strong consistency. They're like a team that always makes decisions together. 2. Apache Cassandra: Uses a different approach called "eventual consistency." It's more like team members working independently and syncing up later. Pro: Super fast writes and high availability Con: Might temporarily have inconsistent data across nodes 3. MongoDB: Offers a spectrum of consistency options: In replica sets, it uses a consensus algorithm similar to Raft Also provides eventual consistency options for faster performance 4. CockroachDB: Uses Raft for strong consistency while scaling horizontally. 🤔 Question for you: Have you encountered consistency issues in your distributed systems? How did you tackle them? #DistributedSystems #Engineering #TechLeadership #Raft 🔗 Part 2: https://lnkd.in/dRhU3B63 Join the Journey with Vikas K. 🔥 Follow for a World of Insightful Content!

9

Worth the time to read in depth. https://lnkd.in/edtPp7RE

4

1 Comment

While there are certainly individual variations in skill and experience when discussing modern software development the story is much more nuanced than pure individual effort.  * software development is a team sport -- individual performance is heavily impacted by the system in which they are asked to operate in.  * we pair skill and experience to this model of understanding performance by expecting people to be able to influence and adapt in direct proportion to their seniority, aka the more senior you are the more responsible you are for the outcomes of your work, the less senior you are the more your manager and management chain are responsible.  * humans want to succeed, we're social animals who derive purpose from demonstrating skill, and working collectively. All of which is to say there is no way you have over 1000 people underperforming at your company without senior leadership deeply implicated in.  If nearly 10% of your company is failing, versus just having the market move against you, you need to fire the CEO.

130

8 Comments