Mauro Restuccia

Coming out of AWS Re:Invent last month was an announcement of a deepening of the relationship between AWS and Anthropic. A key component of the announcement was an additional $4B investment, so total of $8B invested by Amazon. The agreement provides Anthropic with access to Trainium accelerator hardware and a partnership with AWS Labs to work on future hardware initiatives. AWS is really spreading it bets out with GenAI. They announced a whole new family of Amazon models called Nova (RIP Titan?) and continue to offer a best-in-class LLM zoo via Bedrock (missing only OpenAI which is locked down by MSFT). MSFT appears to be pretty tightly coupled to OpenAI (for now) while Google Vertex has a similar LLM gateway solution. I think the biggest gap in the AWS offering is a pre-packaged Enterprise solution similar to the ChatGPT Enterprise offering being used by JPMC and others. The Sagemaker and Bedrock solutions are still DIY for Enterprise customers, great for flexibility but missing some of the speed to market of the pre-packaged offerings. Amazon Lex feels largely abandoned. The home page features videos from 2021 and 2022. That's forever ago in AI years. Where is the Nova GPT store? Thoughts? #AWS #Amazon #GenAI

13

If you're a Director/VP or CTO running an engineering org and will be in NYC Tuesday night (11/19) apply to join us as we discuss personal productivity. Whether it's motivation, organization or optimization, we're going to dig into how we deliver more value, and better value what we deliver :) Should be a great discussion! Link in the comments!

28

2 Comments

Looking to become a better leader? Check out these 10 tips to build trust and propel your team to success! - Authenticity is key: Share your strengths and areas of growth. - Empathy matters: Seek to understand your team's perspectives. - Walk the talk: Align your actions with your words. - Invite honest feedback: Create an open door for communication. - Follow through: Promises made, promises kept. - Respect confidentiality: Build a safe space to share. - Own your mistakes: Sincere apologies foster accountability. - Know your limits: Collaborate and learn together. - Respect boundaries: Promote work-life balance. - Kindness is power: Strengthen relationships, enhance trust. Click to find out more in my latest blog post! 👇 https://lnkd.in/gzm6QYBq

2

Our latest on MongoDB below

2

I had the opportunity to discuss Verto’s approach to roadmap management with Eric Ignasik and Michael Sols from The Software House for the "CTO vs Status Quo" interviews series 🤝 Our main criteria for feature evaluation include: 1️⃣ Any new initiative must increase operational efficiency 2️⃣ The initiative must serve our enterprise and MNC customers 3️⃣ New development products/features must clearly impact revenue & active users — our North Star metrics. And often, I ask my team to actually draw out the UI for what they propose. All the flawed assumptions then come out, and greatness can shine. I hope you will find the interview interesting to read. ➡️ https://lnkd.in/eTPg9a3G #softwaredevelopment #cto #softwaredelivery

80

3 Comments

🤔 Interesting view from a Founder in my network, I know some will disagree. Their A Tier Stack: “Frontend Framework: Next.js Solid performance, great developer experience, and seamless integration with... Deployment: Vercel AWS Amplify is powerful, but Vercel's user experience is just chef's kiss. Deployments are a breeze, and the integration with Next.js is unmatched. Speed and simplicity are key for start-ups, and Vercel nails both. UI Components: Shadcn UI Copy-paste configurable components? Sure. Beautiful design, flexible, and saves a ton of development time. Database: Supabase Out-of-the-box integrations, Row Level Security, and an intuitive UI make this a winner for us. Gets your database up and running quickly and securely. Authentication: Descope This product blew me away. I looked at every auth provider, and Descope is in a league of its own. Multi-tenancy out-of-the-box, super easy setup, and phenomenal customer support, even for early-stage start-ups. Consent Management: Termly Affordable and makes creating, managing, and hosting all your necessary notices and forms a breeze. SOC 2 Compliance: Drata Getting SOC 2 compliant can feel daunting, but Drata makes it manageable. Competitive pricing, great service, automation that saves time, and even instalment payment options – perfect for start-ups watching their cash flow.”

3

2 Comments

I got an opportunity to speak with the power duo from MongoDB - Tara Hernandez and Stacey King Poling about Developer productivity at MongoDB and how should teams think about setting up a DevProd org. Tara and Stacey are some of the most experienced leaders in developer productivity. The whole episode is full of powerful nuggets. Listen to full episode on Substack, Spotify, YouTube or Apple podcasts: https://lnkd.in/gFJD_UHV Some key learnings 👇 - Metrics should be used with caution and in context, as they can be easily misinterpreted or incentivize the wrong behaviors. - Surveys are valuable for understanding developer sentiment, but qualitative data and feedback should also be considered. - The focus should be on outcomes and meeting goals, and managers should prioritize deliverables and create a positive culture. - Building a developer productivity organization requires understanding the business value, educating oneself, and getting buy-in from senior leaders and engineers. Early investment in developer productivity and building good habits is crucial for scaling a company's engineering team. - To change the culture internally, start by identifying and improving one specific area that needs attention. #developerproductivity #developerexperience

39

3 Comments

In two minutes, Zuck explains how scaling works and investments will continue to 10x until scaling stops working. If it ever does! Here's my full analysis of whether we can continue scaling toward AGI: https://lnkd.in/gHj4CAM6 #agi #scalinglaws #aisafety #nvidia #genai

26

6 Comments

“I don’t want to be a manager. I want to be a leader.” — Daniela da Cruz, Director, Engineering at Datadog In the latest feature from Built In, da Cruz shares her vision for building an open and collaborative engineering team with a passion for learning and innovation. Dive into the article to get the full story. #Hiring #NowHiring

9

ICYMI: Thoughtworks was at Product-Led Summit NYC! Attendees heard from Joel Zien, Head of Strategic Growth at Thoughtworks as a speaker and the chair, and Carl Nygard, Chief Technology Officer at Thoughtworks, wrapped up the conference with a Generative AI panel. In Joel’s fireside chat with Matt Skavenski, SVP Engineering at Peloton Interactive, they explored why 'hardware is so hard', and drew from years of experience at the crossroads of hardware and software. Curious about some of the universal lessons in product innovation when it comes to software and hardware? Keep reading! Product Marketing Alliance #ProductLedSummit #ProductThinking #BuildBetterProducts

30

Exciting recap from our time at Product-Led Summit NYC! Check out some of the fireside chat takeaways below. ⬇ #ProductLedSummit #ProductThinking

2

As in a great play, #Redis closes this turbulent year with the return of its old-new star. Redis creator Salvatore Sanfilippo (a.k.a. antirez) is returning to Redis Labs, to serve as an evangelist to “bridge between the company and the community”. After keeping a low profile for over four years, during which Redis Labs shifted away from the open-source model he championed for over a decade, it was interesting to read his post. TL;DR, he doesn't have huge issues with taking Redis core off open source. And no mention of the Valkey fork. Can antirez win back the heart of the community for Redis? Read his post here: https://lnkd.in/dDFUbzt5 * For background on Redis relicensing, and the Valkey fork, see https://lnkd.in/d_sr_yYx

10

1 Comment

Shameless rip from Austin Parker’s BlueSky. But this whole thread is just too good! If you are passionate about #observability you’ll love this! Find the rest here! https://lnkd.in/gsvS_Jpz “”” 1. Metrics are good, actually, but most of you use them very poorly. … 4. there are no silver bullets when it comes to telemetry. you're trading between speed of access, data resolution, and retention … 5. you are not going to build a better datadog on-prem, they have a twelve year and twenty billion dollar head start … 18. the actual goal of observability is to have the data you need to find the answer rather than to have the data that contains the answer … 32. if you have thousands of dashboards, you have no useful ones … 44. single panes of glass are for glass cleaner, not o11y “””

40

5 Comments

Building flashy LLM app demos with 60-70% accuracy is easy. 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗶𝗻𝗴 𝘁𝗵𝗲𝗺 𝘁𝗼 𝟵𝟬-𝟭𝟬𝟬% 𝗶𝘀 𝘁𝗵𝗲 𝗿𝗲𝗮𝗹 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲. There's a massive chasm between these low-accuracy demos and production-ready applications that actually deliver value. Otherwise, 60-70 % accuracy is not enough for an enterprise org. Teams that solely rely on developer based manual prompt validation is unlikely to meet the expected quality bar needed for production. The toolset you need to auto-optimize the accuracy of LLM/Agens system: - Ability to do full testing using real+synthetic data - Ability to improve based on the user feedback - Ability to monitor LLM based solution in prod and use learnings to improve LLM/Agents without robust evaluation, monitoring, and feedback loops will get crushed by those that do. In the long run - its is survival of the fittest in the AI world 💪

24

3 Comments

It’s crunch time for major initiatives, especially those involving new or existing design systems. Whether you’re working on digital transformation or AI feature development, aligning creative, product, and tech goals is a massive challenge. With over a decade of experience consulting for product and technology teams, I’ve seen what makes projects succeed—or struggle. A critical factor is evaluating the scope and timeline of change upfront. Some important questions to ask yourself ● Do my deadlines align with my current burn rate? Assess your team’s velocity now. You still have a few sprints to gather the right data and set achievable goals. If acceleration is needed, now’s the time to act. ● How do my goals align with my service architecture? For example, does your customer management system have the endpoints to support an upgraded loyalty program? Even if every endpoint isn’t mapped, having a plan for system transitions and scalability is crucial. ● How are my updated designs being implemented? Pixel-perfect implementation from Figma to production isn’t easy. Identifying changes to components and documenting them requires experience and clear processes. Luckily, I know a guy who can help 😉.

12

1 Comment

Last week, I talked to the VP of Engineering at a publicly traded US bank. Our team is consulting them on implementing GraphQL federation, here’s what’s most interesting about the roll-out: Some context around the bank: - 10k+ engineers - 36k API endpoints - 4k microservices - *Substantial* tech budget & annual infrastructure costs Here’s why they want to switch to GraphQL: 1/ They value speed. For them, every request needs to be 10 milliseconds or less. The fact that Grafbase was built performance-first really resonated with their engineering team. 2/ They have a ton of microservices and thousands of REST endpoints. So, they’re struggling not just with discoverability from the consumer perspective, but also with keeping all the endpoints secure. 3/ Because they have so many endpoints, they also have an enormous infrastructure footprint that’s costing them almost $1B per year. Their goal: Save 5-10% infrastructure costs (8-9 figures per year) How realistic is it? I’d say you can even save more money if you go all-in on GraphQL Federation, but only if you commit to this for a long enough period of time. Because here’s what’s going to happen (in their case): In year 1 and 2, infrastructure costs will go up (because they’re adding subgraphs that are talking to their existing microservices). In year 3 or 4, they’ll start to collapse these microservices into *just* subgraphs. And that’s where they’ll see the cost savings. To make this change happen, we are currently helping them work through these two main questions: 1/ How big should the subgraphs be? I told them there’s no absolute number here. But a good rule of thumb is that the traffic shouldn’t surpass 20-30% per subgraph to keep things balanced. Also, similar entities should be grouped together on one subgraph to avoid scattered data across boundaries. If you have credit card information saved on one subgraph, you shouldn’t mix it with other non-sensitive data. 2/ How do we implement a solid authorization strategy? There are tons of authorization strategies (RBAC, ABAC, Zanzibar, …). But the issue for enterprise companies like them is that none of these strategies fully suit their needs. After talking to their engineering team, we built our gateway hooks feature that allows them to write their own authorization strategies using WebAssembly. – There are a lot of moving pieces when rolling out GraphQL at scale, anyone else with experience at the enterprise level here?

53

3 Comments

If you are jumping into Rust and Lambda, you need to be thinking about observability. OTel is where you want to look. And plugging OTel traces into Datadog will turbo-charge your insights. Here's how to get started https://lnkd.in/gSwA5sWX

16

We work hard to make giving as easy as possible, while showcasing the power of DAFs as the smartest way to give. And what can be easier than tracking your client's DAF into the flows and processes you already know and love? Endaoment + Plaid enables just that. Interested in giving a twirl or getting a custom integration for your workflow? Just shoot a message ✉️

9