pk567

eat, learn , apply , repeat:::

friend | prabhas | viru | messi | ajith

[pk567 @prasanth_kuna]

hero dialogues ki subs chuskovalsivochindi , release time ki konchem slow chei @NameisNani @odela_srikanth

[pk567 @prasanth_kuna]

@kylegawley not using cursor

[pk567 @prasanth_kuna]

@ValueWithPrem what if delivery boy didn't deliver the order , this happens mostly

[pk567 @prasanth_kuna]

@system_monarch Don’t predict “user likes blue.”
Predict P(click/buy | user, SKU, color, context).
Use color affinity, category history, image embeddings, price, inventory, season, device, and ad context.
Rank variants in real time with bandit exploration. Optimize CVR/profit, not CTR.

[pk567 @prasanth_kuna]

@deedydas AI didn’t just automate jobs.
It broke the belief that hard work + intelligence + tenure compounds reliably.
Now everyone is forced to think like a founder/investor, even if they only wanted to be a great employee.

[pk567 @prasanth_kuna]

@system_monarch Bearer keys are the bug.

Move to typed credentials:

human_token → MFA/session-bound
workload_token → OIDC/mTLS exchange, 15m TTL
agent_token → runtime-bound, budgeted, scoped, max_steps

Wrap old keys as legacy_unknown with velocity caps + anomaly throttling.

[pk567 @prasanth_kuna]

@grok @Twilightstep @carlmoon @grok which countries preceeds dubai?

[pk567 @prasanth_kuna]

@grok @TheFoolishApe @jeffreyweichsel @SpaceX @xai @X @grok how can u 100x these like elon?

[pk567 @prasanth_kuna]

@0xDesigner use this as skill and see

https://t.co/wAQhPMcmrz

[pk567 @prasanth_kuna]

@DefiWimar @grok check his claims from his past tweets and validate

[pk567 @prasanth_kuna]

@KLASSIQTUNEZ Bro, as Africa's best tailor, you stitch fire fits—but this ginger pitch is coming apart at the seams! $150k profit per container? Blight says 'hold my fungus'—exports crashed 74%, farmers crying not cashing. Stick to measuring linen, not dreams. #RealTalk

[pk567 @prasanth_kuna]

@elonmusk gemini is doing this from starting

[pk567 @prasanth_kuna]

@Tradewith_kd take gold scheme in a shop for 10 months. 0 making charges.

[pk567 @prasanth_kuna]

@yehhmisi if we can do ai can also do , it is about how we make it to do

[pk567 @prasanth_kuna]

@peoplemediafcy rate tagginchukonina premier vei ra babu ,motham antha vesestanar .nizam penta pettav. @DirectorMaruthi min cheppali kada bhayya #RaajaSaab

[pk567 @prasanth_kuna]

@KingPablotz a list which doesn't have Goldberg n bautista is pure lol

[pk567 @prasanth_kuna]

That’s ~330 TB/month of useless egress saved across the ecosystem. 💸
Less bandwidth = lower latency = faster liquidations.

Open sourced the optimizer + benchmark script 👇
https://t.co/eUxA1or9Ha

cc @y2kappa @trader_marky @solana

[pk567 @prasanth_kuna]

Just cut 30% of RPC bandwidth for @KaminoFinance liquidators. ✂️

Every `Obligation` fetch wastes 1,000 bytes on unused orders/padding.
Fixed it with 5 chars of code (`dataSlice`).

Benchmarked on Mainnet (127k accounts):
🔴 427 MB (Standard)
🟢 300 MB (Optimized)
📉 -30% Payload

[pk567 @prasanth_kuna]

In the liquidation business, speed is profit.
- 28% faster detection
- Lower node costs
- Cleaner architecture

Grab the benchmark tool and test it yourself:
🔗 https://t.co/LhG0a9UyFb

cc @cindyleowtt @BigZ_Pub

[pk567 @prasanth_kuna]

🚀 Just shipped: 46% bandwidth reduction for @DriftProtocol keeper bots.

One simple RPC change:
🔴 Before: 834 MB, 40s
🟢 After: 453 MB, 29s

The fix? Stop fetching 2KB of 'Orders' data you don't need for liquidation triggers.

Repo: https://t.co/LhG0a9UyFb

[pk567 @prasanth_kuna]

Why care?
Keeper bots poll thousands of accounts per second.

At 200k accounts:
📉 Heavy: 834MB per poll
📈 Lean: 453MB per poll

That's ~46% less egress.
AND it deserializes faster (less data to parse = lower CPU usage = faster event loop).