Author: Richard Seroter

It’s getting dark by 5pm here. Not my favorite time of year. Fortunately, it’s still 63 degrees F so I can’t complain too much. Do you find yourself working different hours when it’s light early and dark early?

[blog] How Google supports veterans and military families. Good stuff. I think most every large tech company offers useful programs for veterans adjusting to civilian life.

[article] Boost Your Team’s Productivity by Hiring Force Multipliers. Are you building “talent density”? It comes in handy when you’re in a dynamic industry and need folks who can pivot to new challenges quickly.

[blog] ML in Go with a Python sidecar. Great blog post from Eli. He looks at how you might interact with models that don’t support your preferred language.

[blog] the death of the architect. Is big upfront architecture a prison of your own making? This post looks at the thinking behind lightweight design and incremental architecture.

[repo] LLM Prompt Tuning Playbook. Want to get better at prompting? This playbook helps you ask better questions of LLMs, including how to best use system instructions.

[blog] Introducing your new JavaScript package manager: Deno. Keep an eye on this JavaScript runtime. They’re iterating at a good speed.

[paper] LoRA vs Full Fine-tuning: An Illusion of Equivalence. Go deeper into these two approaches that may perform similarly, but have different side effects.

[blog] Go Turns 15. Happy birthday to the Go programming language. It’s still my first choice nowadays when coding up most anything.

[blog] Welcome to the era of Services-as-Software… where the lines between services and software are blurring. The HFS folks point out that companies are looking to software (AI powered) to replace labor provided by professional services companies.

[site] Modern Java. Here’s an online book for learning Java from scratch. It’s updated regularly, and has some great foundational content for learning the language.

[article] Paligemma Performance. DigitalOcean posted a tutorial that shows how to fine-tune this open Google model.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

What a wild week, eh? I’m going my daughter’s school play tonight and looking forward to a sunny weekend. But don’t check out for the weekend until you skim the links below.

[blog] Announcing general availability of Google Vids: Our new AI-powered video creation app for work to help tell stories across your organization. Take a look at this! If you’ve had issues engaging with teammates or customers via text, maybe a video will work better.

[blog] Fast, Secure, and Simple: Istio’s Ambient Mode Reaches General Availability in v1.24. If you dismissed the first generation of service meshes like Istio because they were too complicated, I get it. But now Istio is worth a second look. This sidecar-less mode is much simpler and more scalable.

[blog] Gemini is now accessible from the OpenAI Library. It just got a LOT easier to hot swap models if you started with Open AI’s offerings.

[blog] Empowering Engineers with AI. The Slack team goes into some depth on the types of AI-assisted tools and experiences they’ve delivered to their engineers.

[blog] Late Chunking In Long Context Embedding Models. Don’t go to any parties this weekend and start off by dropping some “late chunking” knowledge on anyone. You have to ease into it.

[blog] Etsy’s Service Platform on Cloud Run cuts deployment time from days to under an hour. It’s faster to lift-and-shift to cloud, but the payoff is rarely awesome. Etsy changed how they built and delivered software, and the results are solid.

[article] Why CX is becoming the new priority for CIOs. I’m glad customer experience is a top priority. This shift will help internal tech teams see themselves as business accelerators, not a cost center.

[article] Datadog ups revenue forecasts after AI growth, sniffs out new federal customer. There’s money in observability! Just make sure that as you store and spend more, you’re getting commiserate value.

[blog] How to deploy and serve multi-host gen AI large open models over GKE. Can you build out a pool of infrastructure that helps serve these giant LLMs that demand 100s of GBs of memory? With Kubernetes, it’s now easier.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Every Friday, I send an internal newsletter that includes a Richard Take about what’s going on in the industry. By Wednesday or Thursday, I usually have a topic chosen. This week, I don’t. It’s time to go walk the dogs and figure it out!

[article] How one CIO navigated the aftermath of a scrapped ERP overhaul. Tough call for a new leader, but I like the courage. Continuing with a failed program doesn’t help anyone but the vendor.

[article] Developer Experience Has Always Been a Profit Story. Sure. It’s naive to believe anything else. Any optimization to tools or work environment is done with the goal of helping people produce more/better output.

[article] A Guide to Handing Off Work Before a Vacation. Good advice here! It’s always a good time to take vacation, but make sure you’re not leaving your co-workers in the lurch.

[blog] Now run your custom code at the edge with the Application Load Balancers. Here’s a very cool capability to deploy plugins at the edge of the Google network to act on request/response traffic.

[article] Do user-centric engineering teams build better products? Good question! This post continues a look at the recent DORA survey and emphasizes how important user centricity is.

[blog] Big Ball of Mud – the world’s most popular software architecture. Good post that looks at how architecture emerges, and a handful of patterns.

[blog] Getting resource recommendations from VPA. When running a Kubernetes environment, you might not WANT automatic scaling. That’s cool. William shows how to get advice, not automated reactions.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Quiet-ish day at the office so I had a chance to tackle a few tech projects from my backlog. I did some code customization in Gemini Code Assist which used private code repos to give me suggestions. And I tried out new functionality in our App Integration service. Good stuff. Also, I read some things.

[blog] Tracing the Line: Understanding Logs vs. Traces. What’s a log and when do you use it? How about a trace? This post distinguishes the two from each other.

[article] How To Fail at Microservices. Good look at the various patterns/anti-patterns for running distributed services in your architecture.

[article] To Help Your Team Learn, Set Them Up for Productive Failure. More failure talk? Sure, it’s a good teacher at times. This post looks at how team leaders can help teams take risks.

[article] Create a Full-Stack App With Go and React. Here’s a solid tutorial that shows some open technologies on the frontend, backend, and in-between.

[article] Platform vs. DevEx teams: What’s the difference? You might have a platform engineering team in place, and are considering a developer experience team. Are they different? Same team with different labels? This post spells it out.

[article] Microsoft Security Weaknesses Creating Huge Opportunity for Google Cloud. No lies detected. Microsoft has some fundamental issues to figure out with regards to security.

[blog] Scaling container resources automatically with VPA. William has a great post about vertical pod autoscaling in Kubernetes. What I really liked was that he let this example bake for a while, and showed the results over time.

[blog] Are SDKs Still Relevant In The AI Age? You could think that AI might generate all your code versus needing pre-built SDKs. But that doesn’t seem like the best idea.

[blog] Your Customers Don’t Care About JavaScript. I have no idea what tech stack sits beneath my bank’s systems, and I don’t really care. I want it to be fast, safe, and available.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Election day! It’s a privilege to get to vote and I assume everyone wants what’s best, even if they have different worldviews on how to get there. The tech world kept humming along today, and I read some insightful pieces that you’ll find below.

[article] The Latest in the World of Web Engineering (Featuring AI). Catch up on a handful of topic related to developing for the web.

[article] Why multi-agent AI tackles complexities LLMs can’t. Agents are the next big thing. We’ll see them baked into services and doing some sophisticated activities on our behalves. This post goes into some depth on it.

[blog] Go 1.23’s Game-Changing unique Package: Supercharge Your Data Deduplication. Go’s standard library continues to impress me. I wasn’t aware of this package, but it seems quite useful for inline de-dupe.

[blog] Learning Go in 2024; From Beginner to Senior. Speaking of Go, I read this excellent breakdown of training suggestions for those trying to achieve different levels of expertise with the language.

[blog] No wait, of course THAT is the single most important SaaS metric. There are many good metrics, but choose the ones that matter right now.

[blog] Deploying OAuth2-Proxy as a Cloud Run sidecar container. A sidecar container lets you add functionality to an app without impacting the main container. Here’s an example of using a sidecar for authentication.

[article] What Every Developer Should Know About GPU Computing. Check this out for a very good breakdown of what GPUs are and how to use them.

[blog] Scaling electric transport optimization with Google Cloud Run Jobs. Cool use case, and some terrific results from using the right tool for the job.

[blog] Automating Infrastructure as Code with Vertex AI. This is primarily a post about how prompt design and a long context window can be a better choice than a sophisticated RAG setup.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Happy Monday. It’s Election Day in the States tomorrow. On Wednesday, I expect the world to continue turning regardless of the outcome. Pass the time by reading some tech content!

[article] Platform Engineering’s Most Critical First Decision. Start from the developer portal and go down, or start from the backend platform and go up? That’s the debate called out here.

[blog] Writing secure Go code. There’s some great built-in and community tooling for validating, scanning, and testing your code for security-related concerns.

[blog] Firebase App Hosting: More regions & manual rollouts. This team is quietly delivering some very interesting IDE services (Project IDX), AI frameworks (Genkit), and modern app hosting environments (Firebase App Hosting).

[article] If You’re Not Refactoring Every Day, You’re Making a Mistake. Always be refactoring and modernizing. Don’t wait for a project; keep improving the components you operate.

[docs] Migrate from AWS to Google Cloud: Migrate from AWS Lambda to Cloud Run. This is a very good guide that helps you upgrade your serverless game.

[blog] Why software only moves forward. Does software ever really rollback? This post looks at why you always have to move forward given how data evolved after you shipped.

[article] OpenAI’s ChatGPT search engine challenges Google and Microsoft, and further upends the web. The much-anticipated search experience is out from Open AI? Can they challenge the king? We shall see. Probably not yet.

[article] Voyager 1 Just Phoned Home From 24 Billion Kilometers Away On A Transmitter Not Used Since 1981. I still consider the Voyagers to be one of humanities greatest achievements. We used to build stuff that lasted! Same with Roman concrete.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

That’s a wrap on the week. Enjoy some time outside this weekend!

[blog] Pushing the frontiers of audio generation. This is an area of massively accelerating progress. Here’s what we’re up to, and it includes some embedded audio samples.

[article] How to Build a Billion-Dollar Marketplace — Do’s and Don’ts from the Growth Expert Behind Grubhub, Pinterest, and More. I thought this was a very helpful article for those building up marketplaces. There might be some counter-intuitive advice in there!

[article] What is the impact of AI adoption on engineering organizations? (2024 DORA Report). This analysis digs into the impact of AI on productivity.

[blog] Transformation as a Project. Marty says that you shouldn’t treat transformation efforts (such as moving from a project model to a product model) as a project in itself. Instead, treat the transformation effort itself as a product with the learning and iterations that come with it.

[blog] State in Infrastructure as Code tools. It’s mostly unavoidable that you’ll have state to manage alongside your IaC tools. It might be info about the resources themselves, history, or even random identifiers. Brian takes a look.

[article] Google Just Announced a Bunch of New AI Features for Maps. Some of these look fairly useful. Here’s our blog post too, which shows visuals.

[blog] An intro to Google Cloud for AWS experts. This is a useful look at what’s different (better?) in Google Cloud and can help you map your existing AWS knowledge over.

[article] How To Have A Happy Retirement: 4 Secrets From Research. I’ve got a few years to go. But I do think about “what’s next” and how to stay enthusiastic about whatever I’m doing.

[blog] Google Cloud database news roundup, October 2024 edition. There’s a lot more than just AI going on. I like seeing so many wide-ranging improvements to cloud database offerings.

[blog] Embeddings are underrated. Cool post that explains the fundamentals of embeddings, and how they can be awesome for even folks in tech writing.

[blog] Cloud CISO Perspectives: 10 ways to make cyber-physical systems more resilient. This edition is stuff with links and specific advice for security folks.

[article] It Literally Pays to Love Your Work. People can tell when folks love what you’ve built, and perceive it as more valuable.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Happy Halloween! I’m looking forward to taxing my kids’ hauls tonight and leaving them all the weirdo candy. There’s some treats (no tricks) below.

[article] AI Code Assistants Are Moving Beyond Auto-Complete: Here’s What’s Next. I suspect that AI is going to change what we expect of IDEs and code editors, and even change when we use them!

[blog] Gemini API and Google AI Studio now offer Grounding with Google Search. Get fresher and more trustworthy answers by grounding your results with Google Search results.

[blog] Make It Ephemeral: Software Should Decay and Lose Data. When storage is nearly free, we (accidentally) save everything and avoid purging. This is a thought-provoking piece about re-thinking that.

[article] 2024 DORA Report. I’m seeing more analysis of this report. This time, its insightful perspective from Laura at DX.

[blog] Ask a Techspert: What’s the difference between a CPU, GPU and TPU? Here’s a quick look at different chip types, with extra attention on TPUs.

[blog] Introducing an industry first: application awareness on Cloud Interconnect. If you’ve got cross-cloud setups, you’ll want to take a look at this.

[blog] How Executives Can Avoid Being Disrupted by Emerging Technologies. Most of this AWS article is about how to best identify trends, which is definitely useful.

[article] Chainguard CEO Warns of Open Source License Erosion. Sounds like this was a good talk, but the article recap gives us the salient points.

[blog] Jumpstart analysis with Crashlytics’ AI Assistance: code context. AI-based interactions for developers are getting more sophisticated. I like this ability to give extra context to the agent to help with troubleshooting.

[article] That which never moved can never move back. Is any repatriation from cloud happening en masse? I don’t think so. Some workloads move around. But nobody who’s scaling or trying to win is doubling down on-premises.

[blog] Speed, scale and reliability: 25 years of Google data-center networking evolution. Not all data centers are the same! Amin talks about some pretty special aspects of the Google data center network.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Another big reading list today. There’s a lot happening right now! Check out items below on AI agents, app modernization, and mastering conflict resolution.

[article] Using the Strangler Fig with Mobile Apps. Can you do an incremental modernization on a mobile app? Sure, why not? ThoughtWorks folks explain it in depth.

[tutorial] Deploy and inference Gemma using Model Garden and Vertex AI GPU-backed endpoints. All these open models are terrific, but you need a way to serve them to users. This new tutorial shows how to offer up Gemma 2 for inference.

[article] Beyond Skills: Unlocking the Full Potential of Data Scientists. This article asks whether you’re using your data scientists’ talent to its fullest.

[blog] An Introduction to Astro—A Web Framework for Content-Driven Websites. So many frameworks. If you’re generating static sites, you might be looking at Astro.

[article] How to Master Conflict Resolution. Do you lean into conflict? Most don’t. I had a manager once tell me that I needed to get more comfortable with conflict. Read this to get better at it.

[blog] Bringing developer choice to Copilot with Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s o1-preview. Model choice comes to GitHub Copilot. Use great options from Anthropic and Google in addition to OpenAI. More news from GitHub Universe.

[blog] Fuzz Testing Go HTTP Services. are you writing fuzz tests and inserting lots of valid, nearly valid, or invalid random data to your app to see what happens? Go has fuzzing baked into the standard library, and this is a great look at how to use it.

[blog] Migrating billions of records: moving our active DNS database while it’s in use. You’ve done a database migration before, yes? If so, many of the ideas in the Cloudflare post will be familiar. It’s a good story of migrating a live datastore.

[article] Salesforce AI chatbot Agentforce hits general availability. You’ll see a lot of these announcements in the coming year. Agents everywhere. It’ll be interesting to see the follow-on effect of having agents working without human intervention.

[article] Tabnine launches its code review agent. Tabnine may go below your radar, but pay attention to this shop that’s helping some of the best engineering teams write better code. More here.

[youtube-video] What are Hugging Face Deep Learning Containers? Wietse does a good job explaining what these containers do. If you’re training, fine-tuning, or running open models, you should give this a listen.

[blog] C4A VMs now GA: Our first custom Arm-based Axion CPU. It’s cool to see this now available. As are sixth-generation TPU chips.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below: