Alex Volkov

A meta note about this talk: I am, as a lot of folks out there know, not exactly bullish on LLMs in general. I think most of the hype is nonsense—even while I think some of their capabilities are really astonishing!—but I do think they’re here to stay. So: First of all, if we’re going to be working in a world where people are authoring code with LLMs (Copilot etc.) and using them to “do things” as agent-based systems, the world of inputs to the models matters, as do the environments in which they run. They matter a *lot*! Second, I think a lot of engineering organizations pretty consistently underinvest in foundations, because of the simple realities that many kinds of foundations projects take longer and have a less direct/measurable connection to profitability. They just do, no way around it. Third, then, what are the ways that engineering leaders—including managers, but especially Staff+ engineers—can mitigate the risks of LLMs and improve the chances that they are net neutral or even positive, rather than negative, for software quality, UX, and business outcomes? Secretly (and yes, I am saying it out loud now!) it is predicated on my belief that THESE ARE THE THINGS WE SHOULD BE DOING ANYWAY… but this is a moment when we have a really obvious reason how and why these things matter, with which we can make that case to decision-makers. And because these are the things we should be doing anyway, an organization which uses this LLM-hype-moment to improve the foundations on which we run LLM-related systems… will be in a better spot two years from now *even if the LLM ecosystem craters*. Seems like a win.

10

I'm very proud of a bunch of new search features we just rolled out at DeepCast. Vera Kalinichenko did a lot work early on to evaluate, select, and prototype the foundation of our search architecture, and Oleg Kozlov did the rest of the work to put it into production. This first version was strictly scoped to searching paragraphs in transcript. It was fast and it worked well. In true MVP fashion, we also added a super naive substring match to also search for podcast titles, and episode titles but that obviously didn't work very well in a lot of situations. Since then, we've done a lot of work to identify and summarize common topics, and we've finally combined all of the information into the same search architecture. It sounds complicated, but TL;DR it just made search a whole lot easier to use. Watch the video below, or even better, head over to https://deepcast.fm and try it out for yourself. At the time of this writing, we've added almost 7 million vectors to the database but results are coming back almost instantaneously. That's a huge testament to the design choices Vera made when she made all of the initial design decisions. #search #features #alwaysgetbetter #drivethemission

36

2 Comments

When I shared that CDNs don't "really" work, because the web is becoming so dynamic, personalized and AI-native, I never expected such a widespread response. I've been hearing this sentiment from CTOs and CDOs that tell me they spend way too much on a technology stack that's letting them down, but it turns out it impacts everyone. Not to mention developers hate the duct-taped caching solutions around their applications. No local developer experience, no way to roll back application and CDN atomically, lots of fear, lots of slow client side JS, lots of spinners, lots of hacks. Here's how we can move past CDNs and IaC into a world that's more automated, more efficient, and astronomically better for customers: https://lnkd.in/gv7S7cge

199

13 Comments

New blog post: "Subtitle Shenanigans in SwiftUI’s Menu" SwiftUI’s Menu component is a delightfully powerful API in terms of the complexity of the interactions it supports compared to the API footprint, but if you need to use subtitles then things can go a bit off the rails. Let’s dive in to Menu use cases to get a better understanding of its sometimes odd behavior: https://lnkd.in/d-Pm4kMB

6

2 Comments

I've just deployed my first space on HuggingFace! This space is using a specific T5 that was trained for topic classification. I'm just amazed at the speed that one can deploy a streamlit app - without any docker and minimal configurations thanks to HuggingFace! I also minimize code duplication by installting my main repo and slapping its url in requirements.txt Next step - the inference is pretty slow - 280M parameters inference is not a walk in the park. I think I need to tweak some configurations to improve on this https://lnkd.in/dWKScD5U

13

Abstractions leak in the worst case. So it's always good the low level details behind the abstraction. Despite this, abstractions are super important: What if there was no abstraction in computer networks. Imagine every backend developer implementing TCP in their web server. It would be total chaos!

8

2 Comments

Enums enhance code readability and clarity. They define named values for model attributes, making them ideal for representing statuses, stages, and states. An article from the team at reinteractive explores the powerful features of enums, particularly how they make code readable and self-explanatory … https://buff.ly/4cKGWiu #RubyOnRails #Programming #Enums #reinteractive

1

Just like the big boys, bookend.ai now supports Llama3, alongside our menagerie of open source LLMs. While Llama3 is great, it's also just the tip of the iceberg. What will truly separate the herd from the herd-wranglers is going to take a lot more work than house training a llama. Real value lies in automating the real-life, workhorse apps that power the enterprise and that requires maximizing both safety and simplicity without compromising accuracy and flexibility.

16

At the core of effective engineering leadership is the ability to stay connected with your team. This is perhaps the most challenging aspect because retaining the respect of your people is difficult unless you remain technically involved yourself. In my journey as an engineering leader, staying true to the individual contributor role that originally earned their respect has kept me close. This approach extends beyond technical expertise; it's about maintaining a deep connection with the core activities that propel product development forward at Gearflow. One of my biggest professional fears is becoming irrelevant or disconnected from my team and our work. This fear motivates me to continually reinvent and retool in response to market demands. Next month, I'll be sharing about my adventures generating synthetic data recently. And while I'm no expert when I start preparing for these talks, the process is incredibly valuable. It's less about delivering a perfect talk and more about how the preparation expands my thinking and exposes me to new capabilities. This helps me connect the dots to opportunities at Gearflow that bring real innovation to our customers. Speaking regularly holds me accountable and pushes me to learn ahead of the curve, ensuring I remain relevant and impactful day-to-day. Every talk is a journey of discovery, helping me to lead with humility, communicate clearly, and inspire a culture of innovation through a deep, technical understanding of our work.

20

Hey folks! I'd like to more formally introduce and talk about a few of the projects we're working on towards improving large-scale bioinformatics analyses (one of which was developed during St. Jude's KIDS24 Biohackathon)! If you're not aware, my team started a project called St. Jude Rust Labs (https://lnkd.in/eXY_9dJ8) where we're rewriting a more foundation for bioinformatics analysis in Rust. This cuts across more than just workflow execution, but, recently, we've been primarily focused on improving the experience of working with the Workflow Definition Language (WDL) at scale. To that end, we've released a number of projects, including: - We've written a complete lexer/parser for WDL v1.2 language (https://lnkd.in/eGWHSM2g). Anyone can use this foundation to build tools for WDL. - We've written a VSCode plugin that includes a language server protocol (LSP) for linting and validation directly in your editor (https://lnkd.in/euApG9tF). - And, announcing today, we've started down the road of writing our own WDL execution engine spread across the Crankshaft (https://lnkd.in/et_RFkA3) and Sprocket (https://lnkd.in/eUTuPE68) projects. Crankshaft was prototyped during the St. Jude Biohackathon last week—it's a _headless_ workflow execution engine, meaning that, in theory, others could come along and write drivers for Crankshaft built on NextFlow, CWL, Snakemake, etc. As I said earlier, our team is really focused on WDL specifically, so we're going to continue building out a "head unit" for Crankshaft using WDL. That being said, I would love to see other community projects popping up and using the core machinery for other workflow languages. Thanks to all of the individuals who participated on our Biohackathon team (Kevin Benton, suchitra chavan, Braden Everson, Andrew Frantz, Michael Gattas, Peter Huene, and John McGuigan)!

76

1 Comment

Those of you out there using Claude, make sure to enable the new artifacts feature (still in preview). It's super useful, it can even under certain circumstances run code it generates so you can preview the results. Try it!

19

4 Comments

Learning to Rank is Back on the (Reddit, Inc. Eng Blog) Menu Boys!! In this post by Chris Fournier we discuss some of the infra challenges for Learning to Rank in Solr

62

In this episode, I interviewed Paul Ruvolo, the developer of the CLEW app, about a new app under development. It may or may not be called Navishare, but like the app CLEW, it will allow users to map routes and spaces. Unlike CLEW, it will use AI to "stitch" different points together and allow these routes and maps to be shared. I'm excited to learn more about it and hope it continues development! Clew App on iOS https://lnkd.in/eKe-frm2 https://lnkd.in/eYnTQjSd #Accessibility #AIForGood #AssistiveTechnology

1

🌟 Big news! Amplication AI is now on Product Hunt and we need your support. Jovu is ready to change how developers build apps. If you’re excited about making coding easier and more efficient, support us and spread the word! Visit us here: https://lnkd.in/dDtzUQCg and check Jovu yourself (link at first comment) Amplication #productlaunch #aicodegeneration

30

13 Comments

TL;DR Upstream is June 5! The agenda is now out, and you should sign up, it's free! Long version: It's 2024, and, last I checked, the amazing open source ecosystem we rely on is struggling. 😥 Corporations feast on freely available open source code 🍖 , while expecting volunteer open source maintainers to foot the bill of keeping it secure and well maintained. 💸 ☝ 👉 👇 👈 We've created a vulnerability identification industrial complex 🏭 that is overwhelming development teams with extra work and false positives 📈 and seems more focused on security theater than reducing actual risk. 📉 Meanwhile over 60% of open source maintainers have either quit or considered quitting their work because they are burnt out 🤬, underpaid 🎱 , and overwhelmed 🤕, putting the health and future of the code we depend on at risk. 🆘 Against this bleak backdrop, I come bearing hope! 😻 If you ❤️ open source, if you rely on it in your work, if you make it yourself, or you just care about its future 👉come join us on June 5th for Upstream👈, a free one-day virtual event celebrating open source, the people who use it, and the people who make it. 🎉 This year we'll be exploring a series of exciting new ideas for how to make the open source ecosystem we all depend on work better for everyone. A taste: Vincent Danen, VP of Product Security at Red Hat, will make the case that our current vulnerability patch management strategy is focused on the wrong goal: getting to zero vulnerabilities instead of reducing actual risk (and he'll share how we can fix it!). Jack Cable and Aeva Black from the US Cybersecurity and Infrastructure Security Agency will explain why we need to rethink how security is designed into our technology products and, using lessons learned from how we made automobiles safer, will show us how we can improve security outcomes proactively instead of bolting on security as an aftermarket feature. Frank Nagle from Harvard Business School will share highlights from his recent research that found that our shared open source infrastructure is worth 8.8 trillion dollars 😳 , and discuss with Tidelift co-founder Luis Villa some ways we might preserve or expand that value. There's much, much more, but you should to go check out the full agenda for yourself and sign up to attend. 🔎 https://lnkd.in/g7qwH8wm (s/o to Forrest Brazeal for the image to best my 1000 words)

39

5 Comments

OpenAI released a new feature for its chat completion models where it can be strict about returning exactly the JSON schema that you want. https://lnkd.in/gkFkhprE I’ve share some ideas on how to use the HTTP module in Make to access this new response_format mode, until the OpenAI module is upgraded. https://lnkd.in/gxE_htbr

4

2 Comments

Since this talk from 2011 - not doing a significant percentage of your refactoring via automation - is an outdated way of working when it comes to improving code quality in a product. Refactoring by module or file or function, in every team, is still relevant as part of eg TDD and other flows, but certain forms of code modernization for higher quality and better security can gain a lot of momentum and efficiency by being automation, in any larger scale or safety/security critical software project. This includes - of course - not only the support provided by third party tools like clang, but also tons of own, product specific needs.

1

It's out! Here's my chat on ruby concurrency with the Ruby Rogues. We talk about: - Orchestrating concurrency 🎼 - Colorless programming 💎 - How to approach breaking down concurrent problems 👷🏼‍♂️ - Why understanding concurrency is valuable 💭 - Innovations in the ruby concurrency space 📈 You can find out more here https://lnkd.in/eUV6wtKS or listen on your favorite podcast app!

16