Raghav Agrawal

@impactology
1.4K Followers
3K Following
18.9K Posts
Former Diamond Dealer | Turning research on learning, cognition, and systems into interfaces that reduce confusion, cost and decision stress.

AI agents promised faster delivery. They also delivered 39% more Cognitive Complexity and 30% more static analysis warnings.

A reinforcing loop: more AI code -> complexity rises -> debt accumulates -> velocity drops -> teams write even more AI code to compensate.

Speed gains? Gone within months. Not a tooling problem, a systems problem.

#AIinSoftwareDevelopment #BuildInPublic

Epicure, A multilingual food model trained on 4 million recipes and 7 languages

https://mer.vin/2026/05/epicure-food-ai-4-million-recipes-compressed-into-a-2mb-ingredient-map/

https://arxiv.org/abs/2605.22391

https://huggingface.co/spaces/Kaikaku/epicure-explorer

This is so cool!, you know I wish there was a cool UI tool to create recipes for datasets, like something that even a layman can use to generate or curate or find data and then tune it for their goals, all without knowing anything in data science/data engineering/data analytics or ML

How to Buy Cheap Claude Tokens in China - by Zilan Qian

https://www.chinatalk.media/p/how-to-buy-cheap-claude-tokens-in

This is genius! Chinese people are amazing in chopping up the supply chain

Does a book like this exist? A workbook for product designers who want to make cloud-native and AI infrastructure legible without becoming SREs, Kubernetes experts, or ML platform engineers.

Workbook for practicing the visual grammar/primitives of operational judgment: latency, traces, alerts, dependencies, failure, blast radius, policy, permissions, deployment risk, model behavior, agent actions, cost, uncertainty, and human override.

RE: https://mastodon.social/@impactology/116651799572987810

This is a super interesting thread. I have long been using the Joel Fuhrman #EatToLive book as the template for weight loss. It has basically the same mechanism as the GLP-1 drugs. It is a way to encourage the mechanism described in the below thread. You eat mostly vegetables and mostly raw. You minimize carbs and the more simple the less you want them.

Prior to Ozempic et al, people either didn't believe I could lose weight at the rate I did or couldn't believe it was healthy. When I hold to the plan and can avoid snacking on the many tasty treats littered in this house, I lose on average about 0.6 lbs a day. This is over a long period. At the beginning you lose a lot more because you limit sodium. If I return to the plan after a period of eating like a typical fat American, it's not uncommon to lose 2 or 3 pounds a day at the beginning as I pee out a lot of water weight.

The book Eat to Live is really about 2 chapters you really need to lose weight, a lot of bullshit case studies and anecdotes then a bunch of recipes. In my case, I'm a weirdo who can eat essentially the same salad every single day without a problem. In fact it is easier for me to stay on the plan eating the identical thing than mixing it up. Variety requires planning and thinking. My method requires going to Aldi every three days and buying the same spinach, cucumber, jalapeƱo, bell peppers, tomato and hummus. I make the salad, eat about 1/3 each night.

I don't eat anything of any kind after 9 PM. I try to keep most of my eating between noon and 8 PM so each day I have 16 or more hours of draining the tank. I try to sleep well and drink plenty of water so all the processes that remove fat can work.

After 10 years of going on and off the plan, I am 100% certain it works for me. The only variable is my will to stick with it. When I go off for a holiday or occasion, sometimes I stay off for months or years. When I do my weight goes up. When I get back on, it stays down.

One of the weaknesses is that when I hit my target weight, I had no good end game. I did it, stayed at the weight and enjoyed wearing medium shirts and skinny jeans. At some point, I'd go off for some time period. The longer off, the harder to get back down to my goal weight.

I'm currently on one of these. Early April, after all spring break travels, I got back on the plan. I was 233 lbs. Today, I am 209 lbs. My target weight is 175 lbs. My hope is to get there in early August but I am not putting any real pressure on the timetable. Nice to have, not doing anything drastic to achieve it.

When (not if) I reach target weight, my lifetime plan is this: Monday through Thursday, I eat as I do now. Every dinner is a salad of similar or identical configuration. Friday through Sunday, I can eat anything I want without restriction. Red Rob burger, Brazillian steak house, peanut butter sundae at Brusters, sure. The one catch is that when I weigh in on Friday morning, if I weigh 185.0 lbs or above, free weekend is canceled. This solves the "What, I never get a hamburger" problem but also has a mechanism to keep my weight from growing out of control, to solve the recidivism and yo-yoing caused by lapsing and then restarting.

I can't guarantee this will do exactly what I want but I am hopeful it will work better than what has in the past. I have tubs of size medium t-shirts and size 30 pants I want to get back in so I can't wait to get there.

#weightloss #weightjourney #fat

Visualizing Financial Data by Julie Rodriguez, Piotr Kaczmarek

https://www.amazon.in/Visualizing-Financial-Data-Julie-Rodriguez-ebook/dp/B01EMDJVLO

References

When somebody loses weight, where does the fat go?

https://www.researchgate.net/publication/269772876_When_somebody_loses_weight_where_does_the_fat_go

Molecular mechanisms regulating hormone-sensitive lipase and lipolysis

https://pubmed.ncbi.nlm.nih.gov/14641008/

Mechanisms underlying regional differences in lipolysis in human adipose tissue
https://pubmed.ncbi.nlm.nih.gov/2503539/

Regional and gender variations in adipose tissue lipolysis in response to weight loss https://www.jlr.org/article/S0022-2275(20)33402-7/fulltext

Carnitine transport and fatty acid oxidation https://pubmed.ncbi.nlm.nih.gov/26828774/

The deficit, held over weeks, does ~all the work. So track the scale trend, not your watch's calorie guess (it's off by 20-40%), and pick a deficit you can actually live on.

So why does it net out downward Because the burned fat physically leaves your body : exhaled, peed out and can't be re-deposited.

Eating later builds from new food; it can't recall what's gone. The day bobs up and down, but the floor steps down each night. It's a bank balance, not a memory. Rest isn't cheating

This was the question I actually started with: if I eat and sit and sleep between workouts, how does fat loss "stick"?

Answer : it's a sawtooth, not a switch. Meals nudge the tank up; gaps, cardio and sleep pull it down.

The body isn't holding a streak. Each moment just responds to what's in the blood.