Thiago Mendes

Technology Enthusiast

[Thiago Mendes @thiagomctba]

Let me introduce 𝗦𝘆𝘀𝘁𝗼𝗿𝘆: a web tool to draw and animate your software architecture.

You build the diagram, hit play, and watch the request travel through your system. It tells the story: how it works, not just how it looks.

Try it and tell me what you think: https://t.co/rwQ25HNbWQ

[Bruno Borges @brunoborges]

"The value is not that jaz always beats a JVM expert. A specialist who profiles the workload can match it by hand, or beat it. The value is that you do not have to be that specialist, and you do not have to remember to re-tune every time the container is resized."

Great report by @thiagomctba on the Azure Commands Launcher for #Java

https://t.co/JgtSbxOSus

[Thiago Mendes @thiagomctba]

What if the JVM tuned itself to the container size, with no specialist and no manual tuning?

That is the promise of jaz, Microsoft’s Azure Command Launcher for Java. We put it to the test against the defaults, and the result is in the post:

https://t.co/BoAC3zaqr7

#Java #JVM #Azure #Kubernetes #FinOps

[Thiago Mendes @thiagomctba]

Marreta Lang is live

Write REST APIs as business logic, the language handles the plumbing. Databases, cache, queues, tests and OpenAPI built in, zero deps. Lean by design: tiny memory footprint, fast cold start.

https://t.co/HA65fHcuYS

#OpenSource #RESTAPI #Backend #DevTools

[Thiago Mendes @thiagomctba]

Inside the JVM: the engineering behind enterprise performance.                                                               
What makes the JVM the runtime of choice for high-concurrency enterprise applications?

Read the full study: https://t.co/SJySlE7dLZ

#Java #JVM

[nolen @itseieio]

made a hook that adds a bouncing dvd logo to claude code whenever it's thinking

[Bruno Borges @brunoborges]

Here's how @OpenJDK's Valhalla JDK 26 EA performs compared to JDK 25 for project #JairoSVG

// @BrianGoetz @stuartmarks @gsaab

[Bruno Borges @brunoborges]

I'm pretty impressed with @OpenJDK for image manipulation. Here, #JairoSVG is significantly faster than CairoSVG (Python / C).

JairoSVG is ~34% faster overall (1.34× speedup).

No dependencies. Just Java2D + ImageIO.
What's next: test with JDK 26 Valhalla EA

Full details: https://t.co/VZE0f6moBE

[Thiago Mendes @thiagomctba]

@ildivenire @christzolov .NET actually didn't make it into this V2 benchmark. Definitely a candidate for V3!

[Thiago Mendes @thiagomctba]

@vilojona @brunoborges @otaviojava Great feedback! Will update the legend to reflect latency ranges instead of runtime types for better clarity.

[Thiago Mendes @thiagomctba]

@vilojona @brunoborges @otaviojava Quarkus Native is actually in Tier 2 because it outperformed the other native images (10.36ms vs 14-20ms). Tiers are defined by latency ranges, not by runtime type.

[Dimitris Andreadis @dandreadis]

Do you want to implement a high performing MCP server? @Quarkus leads the pack:
https://t.co/D9meiPf7Ut

[Thiago Mendes @thiagomctba]

39.9 Million requests later. The MCP Benchmark V2 is LIVE.

You asked for high concurrency, GraalVM Native Images, and real IO workloads. I tested 15 different implementations.

Here is what the data revealed.
https://t.co/ntHjCtQO8Q

#AI #MCP #ModelContextProtocol #Benchmark

[Thiago Mendes @thiagomctba]

@brunoborges Thanks for sharing @brunoborges ! ❤️

[Bruno Borges @brunoborges]

If you care about #MCP Server performance and efficiency (cost!) go with Rust, Go, or Java (Quarkus + JVM).

Another amazing MCP Benchmark by Thiago Mendes.

https://t.co/pWWxjEJQaX

[Thiago Mendes @thiagomctba]

@brunoborges V2: https://t.co/ntHjCtQO8Q

[Christian Tzolov🇧🇬🇪🇺🇺🇦 🦋@tzolov.bsky.social @christzolov]

Java is setting the standards for #AI infrastructure⚡
Benchmark of MCP server implementations across 4 major languages puts #MCP #Java & #SpringAI at the top of the performance charts: sub-ms latency, highest throughput, and the best CPU efficiency
📖 https://t.co/kscH4CIKHX

[Thiago Mendes @thiagomctba]

@briancheong @brunoborges Fantastic suggestion, Brian! I will definitely include a high concurrency scenario in the V2 benchmark. Thanks for the insight!