Did very little changes to tsconfig during past 6 months adoption
My day-to-day process - get the new package unless it some antd6, echart or some rendering engine or geo spatial lib, clean up with Claude, strict and unify type system and align it with my vite, tsconfig, oxlint tastes. The result - no need to follow libs bloat and supply chain attack issues. Easy to read, easy to fix.
the original sin of internet - it’s not secure, and for many it’s not the bug it’s a feature to make money or gain power. all nested layers to cover up previous fails. example - nonce, state, encryption bumps in oidc/oauth2.1
Mixed fealings cause the full context should include plans on both Authorization and Authentication flows at least withing Cloudflare ecosystem. No github examples
Anyway good start in the right direction from Cloudflare, yet still long way to go especially compare to the full Ory's offering its built on. Ory's Kratos handles identity, login, registration, recovery, MFA... https://github.com/ory
IMHO full scope should include plans on user store, SAML, multi-tenant org model. Good example - Zitadel https://github.com/zitadel has managed UI for orgs multitenancy, OIDC/PKCE supports, etc you can even partial glue RBAC to it
Siding "MCP is dead, Skills forever" what bother me about all of them is planning to plug MCPs and rotate keys ... this start hitting the fan very soon
My expectations to dear fellow humans - more sophisticated personal insults (ex. give me your cute comments), a freudian slips, hidden messages and motives, first viewer experience with the next cool toy from the hype train, sharing all kind of insecurities, heavy f.. word if very dramatic first person experience happened, border line exposure to the insider info, sharing something your corporate HR gestapo wont appreciate but might help another guy on the line, "i knew the guy who actually did it" stories, motivational statement toward my non-native english, etc
real learning- copy paste the content and ask for “critical and constructive feedback and potential false narratives from industry professionals” to get 10x from it
Real tip - find someone who loves outbound, can create a funnel outside of Linkedin or convert traffic from Linkedin to something more reliable and can talk about numbers non-stop for hours.
Ex. I never did more than 1k whatsapp messages with 20% open rate in a month ...
Know a friend who is doing 190k MRR with 12k whatsapp messages open rate 40%-60% (no AI SDRs!, fake avatars, etc) and what to double it next year. All he wants to talk is outbound ... and how it will make rich and how it should cost no more than 20% revenue.
99,999% hates outbound with passion, want to dump on someone else, can't retain SDRs for more than 6 months, etc
> There are already public memos from large companies where leaders tell their staff that any request for headcount has to come with a justification for why an AI system cannot do the job
spot on! at my place - playwright + prompts instead of hiring QA. data analytic guy is gone ... noone is missing him
today's random quotes
- "AI isn't replacing jobs. AI spending is" ...
- "he job market in India has grown 9% in 2025, so far. 53 million in new jobs. I wonder, how many jobs came from U.S. companies being off shored?"
5 trilllion off the global IT bubble funded by VC money taken somewhere else poured into GPUs and data centers
look at number of linkedin profiles in US companies like Accenture in India .... 450 000 + ... feel really bad biggest transfer of head-counts from US, chatgpt just fuelled it
10+ years in Japan. The message here is much deeper from my perspective. “Let’s jump on the call” is not the solution. The guy was stripped off of his face. I love Japan for being human. Small business bar or restaurant with 3 tables. Not everything should be streamlined for a quick call solution… the process was pushed on his head. Google nemawashi decision making process
Did research on accent, pronunciation improvement, phoneme recognition, kaldi ecosystem, etc … nothing really changed in the public domain past few years. There’s no even accurate open source dataset. All self claimedccc manually labelled dataset with 10k+ hours was partly done with automation. Next issue, model models operates in different latent space often with 50ms chunks while pronunciation assessment requires much better accuracy. Just try to say B loud - silent part gathering energy in the lips, loud part, and everything what resonates after. Worst part there are too many ml papers from the last year students or junior phd folks claiming success or fake improvements, etc
The article itself is just a vector projection in 3d space … the actual reality is much complex.
Any comments on pronunciation assessment models are greatly appreciated
same sentiments with an article author - gpt5 looks like a cost-cut initiative.
my personal feeling gpt5-thinking is much faster but doesnt produce the same quality results as o3 which were capable to scan through the code base dump with file names and make correct calls
I do have lot of hardforked packages, and dont want to look back. Everything just works.
If you confused by the naming, start from Oxlint https://oxc.rs/docs/guide/usage/linter Rolldown https://rolldown.rs/
Did very little changes to tsconfig during past 6 months adoption
My day-to-day process - get the new package unless it some antd6, echart or some rendering engine or geo spatial lib, clean up with Claude, strict and unify type system and align it with my vite, tsconfig, oxlint tastes. The result - no need to follow libs bloat and supply chain attack issues. Easy to read, easy to fix.