faangguyindia

Lately, anytime I want to do anything, AI from Anthropic and OpenAI gets in the way; even simply putting a logo where watermarks from free services exist is now impossible.

I am off to DeepSeek Flash now.

A company that can't make a dominating coding model well, I've no confidence in its general purpose intelligence model; as a result, I don't ask Gemini anything.

MacroCodex app, which results in guaranteed weight loss or weight gain outcomes within 2-5 weeks. It's a completely ad-free, subscription-free product.

Don't believe? Read the reviews.

All you need to do is provide it your calorie intake and weight reading each day.

How does it work? Explained here: https://macrocodex.app/knowledge/rethink/adaptive-tdee/

We already have 16,000+ users.

Human greed has no limits; just because you pay for a product doesn't mean your data will not be sold and mined for additional profit, or that dark patterns and ads will not appear.

to fund training a far better and capable model.

There are far too many things requiring subscription these days, clearly many people can't afford it

I rewrote my Haskell project in Rust.

memory usage dropped to half, execution speed is now 2x. Cross deployment is easy from silicon man to x64.

There are many low level optimizations available which i can exploit, compile is super fast for my project so development experience is better.

LOC dropped to half of the original.

Rust made life so much easier :)

i loved haskell but cross platform haskell deployment is still pain

how do we know it's ai's breakthrough and not ai simply stitching together insights it accumulated from chat with people who are asking questions about the small part of it?

Top contributors on multiple subs were banned by reddit's algo, it's very easy to trick their algo into banning almost anyone on their platform.

>Shares of Reddit were down nearly 23% in recent trading, a day after co-founder and CEO Steve Huffman warned in a letter to shareholders that “search referrals were choppy in the quarter, and traffic was more volatile later in the quarter.”

these are going to be bots. Contributions on top subreddits i used to use are down.

I don't think reddit will survive for much longer, it's going to go away just like stackoverflow and reason for this is going to be their increasing hostility towards subreddit specific mod.

Increasingly, difficult system which bans you for life. Most average user can't work past it. Also their "removed by reddit" algo.

I'd suggest betting on its share fall to nothing.

>? I'm selling it for $1million dollars can it meet that standard.

but people have sold terrible codebases for more than a million dollar.

Method is explained here: https://macrocodex.app/knowledge/rethink/adaptive-tdee/

The actual method used in the app is a bit more complex than this, but I think this explanation is good enough to convey the core idea for now. Later, I plan to publish a more detailed write up.

>Also, how much better it is? How do we even know it’s better?

Very simple! If a car salesman tells you that they use space technology in the engine to improve fuel efficiency, how do you know whether they are telling the truth?

You fill the tank with a specific amount of fuel and drive. If the fuel lasts much longer, you know the technology is actually improving efficiency.

The same idea applies to MacroCodex.

When you eat at maintenance you do not gain weight, when you eat above maintenance you gain weight, when you eat below it, you lose weight.

If you set a goal in the app to gain 1 lb per week and consistently eat the calorie target prescribed by the app, then your actual weight gain should match that rate. If it does, you know the maintenance calories calculated by the app are accurate.

You can choose to set "maintain" or "lose" or "gain".

perhaps using ai.

Many people build calorie / macro tracker like OP but they don't build maintenance calorie tracker.

Your app tracks calorie intake, doesn't keep track of total calorie burn of the day (maintenance calorie tracking)

And no, calorie burn from watch is not good enough!

>A 2020 systematic review by Fuller et al. found that consumer wearables often perform poorly at estimating energy expenditure, even when they measure steps and heart rate reasonably well. Across brands like Apple, Garmin, Polar, and Withings, calorie burn estimates frequently showed substantial error, often in the 15–40% range and sometimes exceeding 50%. Wearables can still be useful for tracking trends, but not for precise calorie calculations.

Here's completely free one i built, https://macrocodex.app/ 16,000+ users already.

I had a similar thought while building a visual workout programming system for bodybuilding.

One thing I keep running into is that workout programs are essentially state machines progression rules, deloads, conditional branches, different exercise substitutions, autoregulation etc...

That made me wonder whether a visual programming model would make these systems more approachable. I'm still unconvinced, though. Unlike automation workflows where the graph itself reveals the logic, I'm not sure what the right visual abstraction is for training programs without making them even more overwhelming.

So far i've a workout programming language and bunch of sample programs and an app which can help you execute these workout plans, all free ofc: https://symbiote-studio.macrocodex.app/?builtin=gzclp

>Compensation works only when the error is consistent.

many people have predictable eating pattern.

>It does not solve macro tracking. This doesn't know if the overage is from fat, carbs or protein.

For purpose of weight gain or loss it doesn't matter.

Deficit = TDEE - Intake

>Adjustments are very delayed and noisy. Your body weight fluctuates with water retention, based on plethora of variables like stress, sleep quality, hormone cycles, etc. The feedback loop tracking weightloss with inaccurate calorie and macro tracking delay or fully impede goals.

well macrocodex uses an algorithm which is designed to deal with all these issues and produce accurate maintenance calorie estimate from calorie intake and weight log data, you can find more details here: https://macrocodex.app/knowledge/rethink/adaptive-tdee/

Later i plan to add more details about the algorithm in a dedicated page.

>For example, 1 tablespoon of oil is 120 calories and 14g of fat.

You should not use them, but for some foods they may work. So if you have a person who isn't going to track the traditional way, then tracking something is better than tracking nothing.

Let's say you are consistently undercounting calories. In that case, free app like MacroCodex (which uses adaptive TDEE tracking) can offset that error and estimate your maintenance calories helping you make better adjustments going forward.

Deficit = TDEE - Intake

macrocodex will simply lower your tdee estimate if you are undercounting, then you'll cut calories further and you'll end up in deficit and lose weight.

I I am working on:

Guaranteed weight loss (for overweight) and weight gain for (underweight) people, you'll see clear gain or loss within 2-5 weeks. If you are in a calorie surplus, your weight goes up. If you are in a calorie deficit, your weight goes down. MacroCodex uses that relationship to estimate your true maintenance calories over time: https://macrocodex.app/ This can be used to gain or lose weight. You simply need to eat above your maintenance calorie line to gain weight and below it to lose weight. No ads, no subscriptions, no payments, 16k users already.

If you're interested in how it works: https://macrocodex.app/knowledge/rethink/adaptive-tdee/

Ad networks like meta ads, google ads use residential IPs + small LLM to check landing page of your advertisers for malware and scams. They are already paying millions for this service.

Sometimes their system malfunctions (cough cough) and you get charged for those clicks but you fail to report them as fraudulent as you've no proof as the IPs and useragent all appear normal to ad agencies and advertisers.

if "2-3 requests per second" per author is what you wanna do on $10 server go ahead". My server does 100 RPS on $16 instance

neither you are saving any time, nor money.

>part from that Django is battle-tested and can help bring a stable "product" quite quickly.

this is a myth, you'll not save anytime. Only way you can save time is if you've experience in this but same is true if you write your app from scratch in Go from your learned patterns.

I mostly use Go + SQLite for all the things I used to use Rails, JavaScript, or Python for.

I find python django wastes too much resources, just look at memory usage.

One of my web app backend (go) is serving approx 100 req/s right now and i look at pprof i see it's not bottlenecked by CPU but mostly IO and i love this.

Writing concurrent code in Go is easy, the code i wrote 10yrs ago still compiles with no issue! This is why i am never gonna switch.

My go apps use very little memory, so we can scale to many users for very cheap.

For larger apps i use postgres (why? replication is easy using pgfailover, high demand apps need multiple api servers so it's out of process db like postgres is fine) but most of my web app use HTMX and if we need some reactivity, i use react (simply due to react experience from work)

For our maintenance calorie tracking app, which is free and has no ads, we have to use as few resources as possible as we scale to thousands of users: macrocodex (which figures out maintenance calories from weight and calorie intake). We initially used Haskell.

Later, it became slow and cumbersome to develop in (developing on an Apple Silicon Mac and deploying to x64 is a pain), even though I liked writing Haskell code. I even tried nix and wasted a day on that! I had a choice between OCaml and Rust. I picked Rust and never looked back.

The algorithm serves in 0.1 ms on Rust. In Haskell, it was 0.2 ms, and memory usage was twice that of Rust. There are many optimization possible in Rust which i didn't do (for sake of simplicity) yet i received good performance.

Yeah, I use Docker to compile Rust, but it's pretty fast, much faster than what I had with Haskell, so the developer experience is great.

By switching to Rust, the LOC dropped to half of what we had in Haskell.

project turned out to be successful. It has already produced guaranteed weight loss or weight gain for many people.

So I set out to create an algorithmic workout app, for which I am using Rust and Go. The mobile app is in Flutter.

Programming doesn't cause me pain or feel like work. I build projects in my free time because I enjoy solving problems.

One thing I love about programming is that it forces me to understand ideas at a much deeper level than simply reading about them. Writing code requires both step by step reasoning and high-level abstraction. That combination often leads to better solutions and a clearer understanding of the problem.

I often ask myself, "Would I still do this if I weren't getting paid?" The answer is usually yes. That's why many of my projects are completely free, with no ads or subscriptions.

Yes, but hydration isn’t really a problem for me. I’m a bit of a water purification nerd, I collect water from a natural waterfall, run it through multiple stages of purification, and then start convincing myself it has some magical properties. I don’t drink soda or caffeine, I mostly just drink water.

most common one, monohydrate.

from your link

>Headache pain often appears along with other signs and symptoms of dehydration, including darker-than-usual pee, dry mouth and fatigue

none of these symptoms are true for me.

yes, i always drink plenty of water throughout day and night too, infact when i get stuck on a problem i go for a walk and drink water.

So drinking water part is consistent with or without creatine. I do know the creatine takes a while to build up, so i don't know why it works but it does.

Sometimes I stay up until 1–3 a.m. because I'm working on a problem that's so interesting I can't stop until I make progress.

If I know I'm only going to get a few hours of sleep, I'll take 5–10 g of creatine, drink plenty of water, and go to bed. The next morning, I usually don't wake up with a headache.

If I skip the creatine under the same conditions, I'm much more likely to wake up with a headache after a short night's sleep.

Now it's become a habit, I automatically take creatine whenever I know my sleep duration is going to be much shorter than usual.

I run pgautofailover with 2 replicas and 1 monitor, you can run 2 replicas on equal configuration, though i size primary bigger and monitor node is tiny.

You can run this on $10x2 = $20 per month setup for 2 replicas and 1 monitor node for maybe $2-3.

For most other projects i just use sqlite, backup periodically to s3.

some report (coincidentally i was checking health of my small cluster for an app)

Common application queries average under 4 ms:frequent analytics queries: ~0.9–1.4 ms average common inserts: ~0.4–3.4 ms average the slower recurring reporting query: 62 ms average across 53 calls, 308 ms worst case

Query volume is approximately 2.30 million SQL statements/day (~26.6 statements/sec), based on pg_stat_statements over the last 97.3 days. That includes every SQL statement, not just user-facing requests: BEGIN/COMMIT alone account for ~1.05M/day, analytics inserts for ~522K/day, and HA/monitoring checks for ~118K/day.

Another thing is that people are optimizing code and algorithms with the help of AI. Many of those AI agents and providers train on the traces, so all the optimizations you came up with in benchmarks will end up in training sets.

At some point in the future, someone can pose a similar problem to an AI one that you spent months, weeks, or days optimizing with the help of AI and they'll end up with the same solution as your final result in less than 10 minutes.

So you need to figure out whether you want to solve the problem today or let others solve it and simply query LLM, let's say, 6-12 months later.

I operate an analytics site (pretty big one B2B where client's backend feeds data into our system), and we see tons of traffic originating from northwestern China (Xinjiang) from Shenzhen Tencent Computer Systems Company Limited.

There are also half a dozen other companies from China continuously hammering our clients’ websites.

I was wondering, what's in that cold dessert? Low and behold satellite imaging shows massive datacenter build outs, very cheap solar energy.

Few months ago something happened and the Geo location on data on those IP now shows "Shanghai" or "Shenzhen". A way to cover tracks? But mapping latency still points to fact that nodes behind these IPs are still operating around Xinjaing region

credit:

'You Can't Cheat Time: Finding foes and yourself with latency trilateration' https://youtu.be/_iAffzWxexA HN user: lopoc

Shenzhen vs Xinxiang is hard to do using this technique but Shanghai vs Xinxiang does show difference.

Assuming that China only distills is a huge mistake.

It’s no longer some backward place that does low value copying. Look at companies like ByteDance and Xiaomi.

Chinese companies aren’t just distilling, they’re acquiring data in the same way American companies did by paying people and crawling the internet.

The way I understand it, China has a few large companies that crawl the web at a rapid rate and build corpora. The government essentially wants select few companies to do this and then make the data available to other strategic companies operating within China.

Then there are data aggregators that buy data from apps, websites, and services, as well as systems like OpenRouter or Cursor, where companies can learn from the “traces” of coding agents, chats, and so on.

This massively reduces costs, as smaller companies like DeepSeek don’t have to do their own crawling or acquire data from 100s of websites and coding agents etc....

There are also companies in China that buy American LLM APIs and proxy them to companies within China. So, there could be 10,000+ companies using American AI products, while China logs all of this, understands how they’re being used, and trains on their traces.