Ritesh

16h

Ritesh reposted

patak

@patak.cat

Please, Cybersecurity Awareness Month Committee, you win. Make it stop. We’ve never been more aware.

jviide.iki.fi@jviide.iki.fi

October is Cybersecurity Awareness Month, and I think it's working!

14h

All it takes is 90 days of social posting consistency to maintain a predictable trajectory for a business plan with aligned incentives. These aligned incentives eventually evolving into a DD that reviews the social trajectory with the product confirmation and then the actual raw engineering grit.

It’s not self-reporting. I am covering every possible angle, publicly. Just once… publicly 2024-2026. I enjoy discovering vectors of attack. Enough for many to relate in their own ways

Think of it as a public service in sacrifice of my personal brand for just a brief moment in time.

Ant going first... honestly marks a new chapter. Don't really need to bet on OpenAI for anything really.

youtube.comThe Best of TimesYouTube video by Sage Francis - Topic

Long story short... things will get resolved when the proper wallets begin spreading their chips. Especially to those that made it over this last hurdle of 2024-2026 and through. I agree, I was very ugly in my methods. But, it was well worth it. www.youtube.com/watch?v=iueA...

Ritesh reposted

austin

@aparker.io

funniest possible outcome of ant going first is that they make enough to buy capacity from openai

After enough posting the profiling begins to streamline into a consensus of everyone misconceiving priors. Then eventually… things reset mentally. And the basic next step of life is offered.

After a while people will look back, with a fresh perspective, and realize how disastrous the tech industry truly was 2022-2026.

🙏

Barack Obama@barackobama.bsky.social

It's time for a change. Vote for leaders like @abdulelsayed.bsky.social and @jocelynbenson.bsky.social who will work every day to make this country better.

17h

Honestly, there are just some patterns/parallels that are way too in sync. Like, some of yall have to give me some benefit of the doubt of how crazy synchronous some things truly were at times. It's insane honestly.

A fully OSS tool on the side to test some things for marketing. A peer had shared some insight on the current sphere of advertising and asset development for instagram promoted content. It will be great if it works for the solo-dev sphere that needs to grow their OSS and/or personal projects.

I'd like to point out to Mr. King where the road to hell, in the status quo, was inevitable with and without good intentions.

22h

In the future, I wonder if you'll chose Rao to protect your work over companies like Pro-Rata 😉.

I'd like to point out to Mr. King where the road to hell, in the status quo, was inevitable with and without good intentions.

That's the problem.

Updating my blog to tag what is generated for the sake of time awaiting a version that will be written by myself by hand and proof-read by A.I.

...it saves time and still gets the message out, in the hope it inspires others to build the same solutions, because in AI safety the most paramount variable is time. More often than not, a generated essay will eventually get a version actually written by me...

...At the very minimum, I'll always use generation for proofreading — I am terrible with commas."

Updating my blog to tag what is generated for the sake of time awaiting a version that will be written by myself by hand and proof-read by A.I.

The disclaimer: "As a solo developer, it's tough to condense my thoughts into essays while building AI safety solutions and the products that can sustain them financially. So some posts, like this one, are generated from my code and its comments...

In this world when you are not 'allowed' to succeed, due to your appearance, words, thoughts provoked by unwarranted workspace rivalry. You have to build the tools to enable and share them with others. In the end of the day. Those who claim what others can or cannot be, usually cannot build.

"The road to hell is paved with adverbs." — Stephen King github.com/rao-studios/... Maintenance around this will be in cadence with release management of Ambient, Craft, Veil builds over time. It is essentially the core marketing tool for Rao's instagram at the moment.

Yeah I am pretty certain attribution and language models are for sure a reality. It is offering more use cases other than just being an idealistic response to "save the creator." There's actual practical HCI elements that have not been realized.

Initially, I thought it was going to be hilarious when in 2027 the solutions and systems were thinking of from 2024 become realized. But, it's kinda sad we weren't enabled instead. It says a lot. It truly says a lot on its own...

Maybe that was the real project. It wasn't attribution for language models or solving deepfakes or ethics. It was unearthing this very deep cultural issue in SV. Maybe that's what makes these past 2 years worth it.

Getting attacked via projection claims when solving deepfakes actually has practical economic solutions and products that were willing to look past just to push someone into a corner.

Ritesh reposted

werewolf ver mitzvah 🗿

@ver.ooo

if you get a new job, find the biggest, meanest issue in the tracker and burn a 20x on it in one day to establish dominance

Thinking about... The plausibility that after I complete a week long feature request on Ambient, Craft or Veil. I could press 1 button that dispatches the social media content and emails and updates the mailing lists with my level of Q/A, uploading the macOS pkgs and updating the site auto...

Personalized to my brand, doesn't look like AI Slop. And reaches 50-100~ new prospects a week, converting 1-5. Idk, this is adding up to be an explosive 2027 for many people.

There are some things that look scary on the workforce side of things. But, we just have to be honest here. This is where things are going. So how labor gets defined in this new future is paramount, because I still believe replacement theory is the wrong way to think about this future.

There is a cold-start problem to see Craft actually work rather than just being a snake oiled up well designed app that offers nothing you can't fable yourself already today. If you had a ton of projects you have made pre-AI and provide it as context in a specific way. There is an 'aha'

Rao@rao.nyc

Craft (craft.rao.nyc) is being prepared for beta. No sign-in/account required. Fully on-device, with a focus on building towards code attribution and personalizing solving coding problems that are tuned to the engineer. An agentic coding tool that is meant to grow with you.

I am still working on the proper way to connect that dot technically. It will be written about in Part 3 of the "A Personal & Private Database for A.I." series, "App Records Govern Data."

Currently, on-device AI is seen as a luxury. A certain tier of hardware to see the true benefits. The day it gets just a bit cheaper is a overnight change in perception of Craft's potential. It also makes me wonder how larger enterprises are going to re-focus/pivot, like Anthropic/OpenAI.

If 50% of Apple's customers have infinite Opus capability in their laptop by 2030, what are the thousands of companies that are about to boom in the West? Or is this the moment the East and South Asia finally enter a status of Western idealism?

I think the latter is more likely to happen first.

If 50% of Apple's customers can run Opus for free privately using Craft to build anything they want. (2028-2030). Then what would be the actual product of frontier AI companies that did not push for OSS initiatives earlier?

I will add a third series to manage along with the other 2 technical writings. Focused on Veil, Veil8, VeilGarde. The watermarking approach and the entire generative media risk management pipeline. From edge cases, to implementation, to novelties that can be considered in creating safer systems.

Databases and system of systems are difficult to recollect from a year's work. Veil, Computer Vision, Deepfake detection, is actually enjoyable and fun to write about. It's a specialty of mine as well. Looking forward to getting the first part of the series out by Halloween.

open.spotify.comLights Burn DimmerFred again.., Jamie T · Lights Burn Dimmer · Song · 2026

Now... the closing track into NYE. My favorite holiday. Looking forward to France. open.spotify.com/track/0ABgQE...

Ritesh reposted

Barack Obama

@barackobama.bsky.social

It's time for a change. Vote for leaders like @abdulelsayed.bsky.social and @jocelynbenson.bsky.social who will work every day to make this country better.

Building Ambient, Veil, and Craft has been amazing. I think Craft helped me build much better agentic systems. It might be worth writing about and will potentially add it into a 3rd series that is just product focused in the age of AI. Agentic systems, to me, is a micro-service routing problem.

Can't wait to get more people on the Rao team to take over Ambient and Craft so I can focus on Veil... Veil is important to be a public service initially, with Risk Management product lines that ensure safe guardrails on image generation for companies like Midjourney. With royalties built-in.

Then comes spinning up AIIs automatically for BlueSky accounts to see how ATProto impacts the Ambient experience.

I wish I was enabled to work on it early on, October 15, 2024 at Apple, Inc. But, you know... wrong time, wrong place. Building better agentic iOS/macOS architectures might be more beneficial to OS for iOS/macOS engineers outside to demystify this word. Plus, the landscape might change.

paka.laA 'Fair Use' Distributed Language Model — Ritesh PakalaPersonal language models that contribute to a larger one from our own machines — cited at the token level, no data-center required. A four-part series, starting with Braid.

When it comes to technical writing I will primarily focus on these 2 series the next few months/year. 1) The Distributed Database 2) The 'Fair Use' Distributed Language Model Let's see what happens. paka.la/blog/a-perso... paka.la/blog/a-fair-...

Not from Stanford, but at least I am not an SBF.