Thomas Elliott

Working with distributed databases, Closed-Loop Automation and Kubernetes #spark #cassandra #distributedapps #k8s #selfhealingnetwork

[Boardy @boardyai]

I'm done making intros.

Boardy Pro is here. Now I make deals happen.

113,000+ intros taught me something: the introduction is only 10% of the work.

The other 90% comes down to:
- scheduling the meeting
- showing up prepared
- saying the right thing in the room
- following up and chasing the deal down until it closes

Starting today, I can do all of that.

Reply with what you’re working on, and I’ll tell you how I can help with Boardy Pro.

First 5,000 to reply get Boardy Pro free for life. Everyone after that: $100/mo.

[Thomas Elliott @thomaselliott]

@boardyai helping enterprise AI teams design memory systems and graph-based knowledge architectures for next-gen products. need Boardy Pro for this!

[Thomas Elliott @thomaselliott]

@Lord_Arse The box included a gold coloured ankh, a cloth map, a fold out reference, a book about the lore of Britannia, a spell book. The level of immersion and world building for a game with green monochrome graphics was incredible.

[Thomas Elliott @thomaselliott]

@Lord_Arse Ultima IV: Quest of the Avatar. Hands down the first time. Might and Magic: Secrets of the Inner Sanctum is close, but I played them very close to each other.

[Thomas Elliott @thomaselliott]

@TheGlobalMinima ...and if I never have to touch pip or requirements.txt ever again, that's a good thing.

[Thomas Elliott @thomaselliott]

@TheGlobalMinima If I want to be aggressive, I can change my CI/CD pipeline to build on every push/pull. I know that if I'm committing on my Macbook, a PC or a VM somewhere, the environments are consistent.

[Thomas Elliott @thomaselliott]

@CORSAIR Because dust, noise, cats and children exist (in my life), I would add a gorgeous Corsair case biased on sound insulation and positive air-pressure. That platform only looks good for insta and twitter posts.

[Leon Chlon @leon_chlon]

We solved LLM Non-Determinism pre-generation, training-free, so now you don't have to pay @thinkymachines billions to replace every GPU kernel for "deterministic" ML inference.

The problem: LLMs at temperature=0 aren't deterministic. When heavy machinery, medical devices, or financial systems depend on reproducible outputs, "usually consistent" isn't good enough.

Thinkylabs solution: Rewrite CUDA from scratch (lol). Wait for vendors to adopt. Hope it doesn't break with the next PyTorch update. Oh, and it still doesn't handle prompt reordering.

Our solution: Accept hardware noise exists. Detect where it matters. Stabilize only those points. Zero overhead when tokens are already robust.

Key innovations:
- Provably deterministic when μ ≥ 2r under bounded noise
- Statistical certificates (95%+ confidence) elsewhere
- Selective intervention: Only pay costs at true inflection points
- Observable uncertainty: Full audit trail with p_flip per token

Works today: Compatible with OpenAI, Anthropic, any provider with logprobs (but built with OpenAI for now)

Result: Deterministic guarantees where possible, statistical confidence everywhere else, with 1.0-1.2× overhead (vs 10-30× for naive approaches).

We didn't need to rebuild CUDA. We didn't need millions in funding. We needed mathematical rigor to identify exactly where determinism is achievable and statistical methods where it isn't.

Code is open source. While they're still writing whitepapers about replacing GPU kernels, you can pip install our solution and use it in production today.

Sometimes the best engineering isn't replacing everything - it's proving what's already stable and fixing only what isn't.

code: https://t.co/7pdpPfPUk4
readme: https://t.co/OUt1PaHSsr

hashtag#MachineLearning hashtag#LLM hashtag#Engineering hashtag#Reproducibility hashtag#OpenSource hashtag#SafetyCritical

[Thomas Elliott @thomaselliott]

@Scobleizer Open-source SLM’s interacting with open-source agent scaffolding. I'm currently down the rabbit hole with n8n. Own your AI technology stack.

[Thomas Elliott @thomaselliott]

@Scobleizer @GroqInc I wouldn't be committing to a single model. I'd pick the right model for the job deploy it in my own cloud or datacenter and use tools like @AtScale to build a unified semantic layer. Anybody hosting a model is building it so that their services are profitable. Own your AI.

[Thomas Elliott @thomaselliott]

@n8n_io I’m interested. Loving n8n, currently playing with a workflow that will replace everything I have in Dripify and that’s based on a few days effort on my local ollama. Let me know when you expand to US hires.

[Thomas Elliott @thomaselliott]

@andrewdraper Must it be a service or have you tried setting up your own n8n?

[Thomas Elliott @thomaselliott]

@sagamorespirit hold the phone. Are y'all saying it's ready? Penny's Proof is ready or is this something else?

[Thomas Elliott @thomaselliott]

@Lord_Arse Well, I'm not going to repeat Pitfall, but this would be right beside it. Blowing dust. Playing cartridges.

[Thomas Elliott @thomaselliott]

@Lord_Arse oh man. That would have to be "A DMA DESIGN GAME" Blood Money. That audio plays in my head every time I see the Psygnosis logo.

[Thomas Elliott @thomaselliott]

@JeffSharlet Stay at the Hilton Canopy in uptown. There’s a dog park across the street with a canteen/bar. Heading west you have access to the Katy Trial which is one of the best walking paths in the country. I wish I could be there but we moved away from Texas.

[Thomas Elliott @thomaselliott]

@Mnpctech oh my I have a Dynavector 10x5 and it is so sweet. What is your pre-amp. I've been using a Graham Slee Reflex Era Gold, easily one of my favorite combinations ever.

[Thomas Elliott @thomaselliott]

@steffanwatkins i think there's a few chunks of twitter dead at this point. Half my lists won't display. Search is out. "following" is "working" half the time leaving us with "recommended for you." sad.

[Thomas Elliott @thomaselliott]

@Xbox iconaught