SpeakEZ.tech's Avatar

SpeakEZ.tech

@speakeztech.bsky.social

Intelligent, Resilient Platform Services for the modern technology landscape #dotnet #fsharp #ai #ml #graph #vector #postgres #linux #iot #cloud #datasecurity #ethicalOSS #ethicalAI #quantum #postquantum #resiliency #sustainability

292 Followers  |  156 Following  |  333 Posts  |  Joined: 17.02.2024  |  1.9423

Latest posts by speakeztech.bsky.social on Bluesky

Preview
Coeffects and Codata in Firefly A Fresh Take on Efficiency and Observability from Storied Mathematical Principles

#fsharp #haskell #scala #rust #ocaml #mlir speakez.tech/blog/coeffec...

03.08.2025 15:43 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
Alloy.Rx: Native Reactivity in Fidelity Porting A Push Model with Principled Performance

#fsharp #cpp #mlir #llvm #AI #iot #webassembly #systemsprogramming #softwaredevelopment speakez.tech/blog/alloyrx...

03.08.2025 15:37 β€” πŸ‘ 3    πŸ” 2    πŸ’¬ 1    πŸ“Œ 0

Thanks! I too hope the Fidelity framework takes on a life of its own similar to the Fable Compiler/ecosystem - independent but with lots of sympathetic "shared edges" with dotnet. There's still work to do for MLIR to really express broad high-level languages but it's a very active ecosystem. πŸ‘

30.07.2025 00:35 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
The Continuation Preservation Paradox How Deep Can Delimited Continuations Survive in a Systems Compiler?

πŸ”¬ #fsharp #mlir #llvm #wasm #embeddedsystems #webassembly #wasm
speakez.tech/blog/the-con...

29.07.2025 14:21 β€” πŸ‘ 7    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0
Preview
Why F# is a Natural Fit for MLIR How functional design can express low-level efficiency with safety

#fsharp #mlir #AI #machinelearning #systemsprogramming speakez.tech/blog/why-fsh...

29.07.2025 03:16 β€” πŸ‘ 7    πŸ” 2    πŸ’¬ 0    πŸ“Œ 1
Preview
Hardware Lessons from LISP Boom, Bust and Broadsides from the first era of AI

#FSharp #OCaml #StandardML #LISP speakez.tech/blog/hardwar...

22.07.2025 15:29 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Hundreds of victims, thousands of lives affected. Even if you don’t believe Epstein’s associates partook in the abuse (which would mean you don’t believe the testimony of his victims), at the very least the story of how the world’s wealthy and powerful enabled the abuse is real and underreported

21.07.2025 13:24 β€” πŸ‘ 206    πŸ” 22    πŸ’¬ 4    πŸ“Œ 0
Preview
Speed & Safety with Inet & Graph Coloring Keeping Parallel Computation on the Beam with a Rational Approach to Formal Methods

πŸ’Έ #AI acceleration should be a commodity, not a monopoly. πŸ’°Our design for the Fidelity framework uses interaction nets to target GPU & CPU/SIMD via #MLIR. No opaque vendor APIs that sacrifice accuracy, efficiency or safety. Just functional code β†’ silicon. #FSharp #LLVM speakez.tech/blog/speed-a...

21.07.2025 02:31 β€” πŸ‘ 7    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0

πŸ˜† to each their own πŸ‘Œ enjoy life!

20.07.2025 11:13 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Try F#

#fsharp ☝️
try.fsharp.org

18.07.2025 20:57 β€” πŸ‘ 7    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Post image 15.07.2025 17:00 β€” πŸ‘ 113    πŸ” 9    πŸ’¬ 4    πŸ“Œ 3

And the runtime diaspora! That's going to take some time to sort out. 🫒

18.07.2025 20:47 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
GitHub - mfp/extprot: extprot: extensible binary protocols for cross-language communication and long-term serialization extprot: extensible binary protocols for cross-language communication and long-term serialization - mfp/extprot

OCaml had already figured out a significant portion of this 17 years ago. They were preserving type information in the wire format when everyone else was bedazzled by shiny versions of corporate surveillance courtesy of Google and Facebook. github.com/mfp/extprot

18.07.2025 19:00 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

On the "Fidelity" name, it was chosen for the metaphor of preserving type information to *not* fall back on heuristics too soon - and - to provide memory layout optimization *to* MLIR rather than waste passes for it to determine its own (and risk getting it wrong).

I just learned this week that...

18.07.2025 18:54 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

It's great you shouted out the F# native markdown processing that's around. The bias here to to break the interop habit wherever possible to keep doors open to port to Fidelity where a component is useful. πŸ‘

18.07.2025 14:26 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Our site is still on Hugo - for now. But of course I want to have some options for middleware, semantic search and other bits and bobs to show a few more of our wares along the way. Baby steps!

18.07.2025 14:02 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

For markdown, we started with the notion of forking Fornax and extending it Hugo-style to deploy to Cloudflare Pages. That has done a full pivot to Oxpecker and leaning heavily into Markdig (for now) with a side quest into presentation formatting using the Fable wrapper for Reveal.js.

18.07.2025 13:29 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

I'm saying the same thing about my own project as I have a talk to give in a little under a month. 😲☺️

Would love to see your work with markdown handling - looks super slick. 🌟

We're building Elmish into the Fidelity - borrowing heavily from @jordanmarr.bsky.social ReactiveElmish and Avalonia/Skia.

18.07.2025 13:08 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 2    πŸ“Œ 0

This is great - just catching up on your project. Have you considered doing an @amplifyingfsharp.io session to showcase bits and pieces you're willing to present? That would be really cool! πŸš€

18.07.2025 12:11 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Preview
Exclusive: Meta fixes bug that could leak users’ AI prompts and generated content The tech giant fixed the security flaw, netting a security researcher $10,000 for privately disclosing the bug.

Move fast and leak things πŸ™„

Meta fixes bug that could leak users' AI prompts and generated content | TechCrunch share.google/WA3goFCVjBy1...

15.07.2025 23:36 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Today, this thought crossed my mind:

Claude Code appears to be the current pinnacle of AI-supported software development. In the videos I've seen, considerable emphasis is placed on the importance of a detailed and precise specification. The better the spec, the better the outcome.

#fsharp

14.07.2025 12:16 β€” πŸ‘ 6    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

It's nice to see models that reward structure in a well designed language. But the issue of more recent API updates needing special prompting yields shorter discourse with the model. I wanted to squeeze F# Compiler Services 43.9.300 for insight and answered my own question in prepping the prompt. ☺️

15.07.2025 01:43 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

The fact that 'bad' F# code is less prevalent in the commons (and therefore not as much mid-code to be inferred due to dilution in training) is part of why things tend to line up well once the prompt orients the model. Another reason is the relatively consistent API.

15.07.2025 01:39 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

πŸ€” "Why are you so focused on low-level memory operations in a functional language?" Here's the thing: We're not abandoning high-level abstractions. We're making them work everywhere 🎯

The path is more direct than you might think πŸ‘‡
speakez.tech/blog/beyond-...

#FunctionalProgramming #FSharp

11.07.2025 19:58 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Microsoft management is a circle jerk of MBAs so why bother learning anything beyond Excel to work there?

11.07.2025 19:49 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Coupling & Cohesion: Structure is Speed Using Functional Programming as a Super-Power for Efficient Compilation

Fun with #fsharp and a hat tip to @rtfeldman.bsky.social for inspo from www.youtube.com/watch?v=vzfy... speakez.tech/blog/couplin...

10.07.2025 12:40 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

We appreciate the clarity (to our view) of the API. We're standardizing on it. πŸ‘

04.07.2025 13:09 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Building Firefly with Alloy How the Fidelity framework achieves zero-allocation performance with F#

speakez.tech/blog/buildin...

04.07.2025 02:41 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Having some #fsharp fun building Alloy's shadow API! 🎭 These look like System/Core calls but they're statically resolved to our dependency-free implementations. Same familiar F# idioms, zero BCL dependencies. It's like BCL cosplay for native compilation! πŸš€ #dotnet

03.07.2025 20:44 β€” πŸ‘ 6    πŸ” 3    πŸ’¬ 1    πŸ“Œ 0

"Copilots! Copilots! Copilots!" #dotnet 🫣

01.07.2025 00:16 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

@speakeztech is following 20 prominent accounts