Igor Eydman

iOS Dev by day, avid documentation and technical book reader by night

[Igor Eydman @IEydman]

Is it ever too soon to ship? Tarpit is taking a bit of time for me to test, but something else might be coming soon. ⛵️

[Igor Eydman @IEydman]

Just shipped SprintVibe at #v0PromptToProduction Miami!

Built with @v0 and deployed on @vercel at The LAB Miami.

Check it out: https://t.co/W8fLgeg1bT

Hosted by @AgnelNieves

#BuildInPublic #MiamiTech #v0

[Igor Eydman @IEydman]

Has anyone had issues getting context from previous sessions in #GPT5 ? It was the easiest way to context prime until recently.

[Igor Eydman @IEydman]

async coding > parallel vibe coding > vibe coding

[Igor Eydman @IEydman]

#WWDC25 Should I update to MacOS 26 beta to use the new fancy pants Xcode beta features?

[Igor Eydman @IEydman]

#WWDC25 For anyone diving into prompting through Apple Intelligence, Apple updated the human interface guidelines https://t.co/dpyPOIBmmM

[Igor Eydman @IEydman]

@rxwei Looking forward to deep dive into this library!

[Igor Eydman @IEydman]

✅ A library for the SLMs to be agentic (Model, Instructions, Tools)
FoundationModels + Adapters Nice!

[Igor Eydman @IEydman]

#WWDC25 is tomorrow and here is my wishlist:
- Small Language Models for devs to use on device
- A library for the SLMs to be agentic (Model, Instructions, Tools)
- A routing library for those SLMs to orchestrate them
- A cloud option to run tasks async
- Agents in Xcode

[Igor Eydman @IEydman]

✅Small Language Models for devs to use on device
Nice!

[Alex Xu @alexxubyte]

Our new book Mobile System Design Interview is available on Amazon!

[Igor Eydman @IEydman]

spoke to soon, looks like it works with podcasts too

[Igor Eydman @IEydman]

#WWDC25 Does workout buddy work with podcasts?

[Sam Altman @sama]

for our next open source project, would it be more useful to do an o3-mini level model that is pretty small but still needs to run on GPUs, or the best phone-sized model we can do?

[Igor Eydman @IEydman]

#tarpitAI hello world.

[Vaibhav (VB) Srivastav @reach_vb]

WOW! Apple just dropped Core ML optimised models for FastVIT, DepthAnything & DETR 🔥

> Quantised models for Image Classification, Monocular Depth Estimation, Semantic Segmentation
> Along with the checkpoints, they report detailed benchmarks on Inference speed, accuracy and more

> Bonus: We ship sample apps to get started with these models quickly!

The time for on-device AI is now! 🤗

[Igor Eydman @IEydman]

With #WWDC24 less than a week away, here is my wishlist:
- On-device inference with a Small Language Model
- SwiftData with vector embeddings
- LoRAs for on-device models
- A framework for multi agent systems
- Local copilot for Xcode

[Greg Brockman @gdb]

We’re releasing GPT-4 — a large multimodal model (image & text in, text out) which is a significant advance in both capability and alignment.

Still limited in many ways, but passes many qualification benchmarks like the bar exam & AP Calculus: https://t.co/L6VGJ0WfFv

[Igor Eydman @IEydman]

2/2 I also think this role would require encyclopedic domain knowledge. Like a living index of stuff related to a specific field or task. So the person can ask the right prompt with proper descriptions and know enough to check the answer to re-engineer the question to fix it.

[Igor Eydman @IEydman]

1/2 With StableDiffusion and ChatGPT being so reliant on good prompts and queries, I’m starting to think that there might be a new career for people that ask good, inquisitive questions. What job title do you think it’ll be? #stablediffusion #ChatGPT #dalle2 #midjourneyAi