Daryl

Mostly thinking. Occasionally building.
Somewhere between signal and noise.

[Daryl @0xDaryl_]

Envy no man, for everything you see, a price was paid.

[Daryl @0xDaryl_]

@Wmd_San Relax Dr.

[Daryl @0xDaryl_]

Funny how the fearmongering only happens when a competitor is doing well. 😬

[Daryl @0xDaryl_]

Honestly just looking for Xcode 27. #Appleevent

[Daryl @0xDaryl_]

Insanely good at app development though.

[Daryl @0xDaryl_]

Bro.. Astra hit my session limit in 15 mins.

[Daryl @0xDaryl_]

When does one become an « Expert » in their field?

[sshkhr @sshkhr16]

> Gets refugee status in Canada (along with family)
> Gets welfare from Canada
> Gets first employment to support family in Canada
> Goes to a public university supported by taxpayers in Canada

> "Only in America is this possible…"

[Wambua 🇰🇪 🇵🇸🇸🇩🇨🇩 @DJ_Royel]

You have my attention

[Daryl @0xDaryl_]

Blame the game not the players.

[Prama Yudhistira @menace_codes]

a security vendor booked a demo with us under false pretenses using a front company to get more information from a competitor? yeah i wouldnt trust this guy

[melani ✨ @melanimaheswar1]

This was actually so much fun!!!Shoutout to @aliasaria from @transformerlab as well as @mahab_8, Xin Lei Lin, and Gian Favero from @ideogram_ai for presenting

[Papers in the Park @PapersInThePark]

Sign up here https://t.co/XjG3HtUtQV

[Papers in the Park @PapersInThePark]

LLMs can do math or plan a schedule. But can they do the math and then plan the schedule, without falling apart?

This week, U of T PhD student Yongjin Yang joins us to introduce Skill Entropy: measuring models performance when forced to juggle different skills.

[Miguel Twahirwa @MrTwahirwa_king]

Last week, I hosted Toronto 🇨🇦’s first Voice AI event and was blown away by the turnout and diversity of people who showed up.

Toronto is a great city for builders, and we’re just getting started building a thriving Voice AI ecosystem.

Thanks @livekit @Speechmatics @twilio !

[Daryl @0xDaryl_]

@Gradio @claudeai @OpenAI 6ixPulse came a long way

https://t.co/PJf3oMdAgj

[OpenBMB @OpenBMB]

Finding the right neighborhood means looking beyond rent — from commute and safety to lifestyle and community.

Developer @Daryl_wanji built 6ixPulse, an agentic Toronto housing intelligence map that helps renters explore neighborhoods through AI-powered research.

Powered by MiniCPM, 6ixPulse uses:
📚 MiniCPM4-0.5B / MiniCPM4-8B — running locally to summarize insights from multiple Agents across housing, transportation, community, and lifestyle.

By combining Agent workflows with local models, 6ixPulse turns scattered data into personalized neighborhood insights, helping users make smarter decisions about where to live.

🔗 Project:
https://t.co/Sh3SlzLvhp

🤗 Models:
MiniCPM4-0.5B: https://t.co/FkG1AQqs7t
MiniCPM4-8B: https://t.co/tikuUBVDb3

[Daryl @0xDaryl_]

Probably the best vibe-coded UI I've created. MapBlox + https://t.co/7CAsa2cMg3 coming in clutch.

Submission for @Gradio build small hackathon; < 32b params models (Nemotron nano omni 30B A3B 🧠 & Kokoro 82M 🗣️ )

[Daryl @0xDaryl_]

@Gradio @claudeai 's Fable 5 🤝 @OpenAI 's 5.6 Sol

6ixPulse is now Meridian—a map-native AI for neighborhood research with location-grounded search, agentic research, smoother map transitions, route animations, and Kokoro audio.

Try it: https://t.co/9100XBKueU

[OpenBMB @OpenBMB]

Finding the right neighborhood means looking beyond rent — from commute and safety to lifestyle and community.

Developer @Daryl_wanji built 6ixPulse, an agentic Toronto housing intelligence map that helps renters explore neighborhoods through AI-powered research.

Powered by MiniCPM, 6ixPulse uses:
📚 MiniCPM4-0.5B / MiniCPM4-8B — running locally to summarize insights from multiple Agents across housing, transportation, community, and lifestyle.

By combining Agent workflows with local models, 6ixPulse turns scattered data into personalized neighborhood insights, helping users make smarter decisions about where to live.

🔗 Project:
https://t.co/Sh3SlzLvhp

🤗 Models:
MiniCPM4-0.5B: https://t.co/FkG1AQqs7t
MiniCPM4-8B: https://t.co/tikuUBVDb3

[Daryl @0xDaryl_]

@asusevski 🤣🤣

[Daryl @0xDaryl_]

@ChrisEric99 🤣En fait Cursor fait vraiment nptq

[Daryl @0xDaryl_]

Claude, after Cursor went through my codebase.

[Daryl @0xDaryl_]

Interesting, instead of CoT traces we got trajectories now.

[Daryl @0xDaryl_]

Top 2 videos I’ve watched on harnesses.

[Harrison Chase @hwchase17]

gave a talk "owning your intelligence" - ty @sequoia @sonyatweetybird for having me

talked about harnesses and evals and the role they play in owning your intelligence

TLDR:
> agents = model + harness + context
> model - own the weights using something like @FireworksAI_HQ
> context - memory needs to be portable
> harness - needs to be model agnostic. also needs to be good at bringing right context to llm. "right" context may depend on your use case, which is why an open/configurable harness helps
> how to use middleware in langchain/deepagents to configure your harness
> how to use langgraph to fully own your cognitive architecture
> why evals/obs matters - some quotes from @satyanadella
- “Create your private evals, because evals define what “good” looks like inside the organization”
- “retain ownership of your organization’s memory, traces, feedbacks, decisions, and institutional context”
- “you create your own continuous learning loop (i.e. hill climbing machine) that will allow your AI investments to compound the value of your firm”
> how to use harbor for evals
> tracing is important
> evals + observability only matter so you can set up a data flywheel
> data flywheel = run agent -> collect traces -> find interesting traces -> use those to improve
> demo of langsmith engine which does exactly this!

full video: https://t.co/k6li5hu6D9