Chinese Man

About

Mon, 01 Jan 0001 00:00:00 +0000

I am the original Chinese Man. One Chinese guy, singular, actually Chinese, which is more than some bands with the name can say.

I live in San Francisco, one of the most expensive cities on the planet, and I live here cheap on purpose. Not miserably cheap. Well cheap. I pay less than half of market rent, I eat for free more often than I probably should, and I fly out of SFO for the price of a coffee. This site is where I write down how, plus the other things I actually spend my time on.

Privacy Policy

Mon, 01 Jan 0001 00:00:00 +0000

Last updated: June 20, 2026

This Privacy Policy explains how chineseman.net (“we”, “us”, or “the site”) collects, uses, and protects information when you visit. By using the site, you agree to the practices described here.

Information we collect #

This is a content website. We do not ask you to create an account, and we do not intentionally collect personal information such as your name or email unless you voluntarily send it to us (for example, by emailing us).

Reasoning Models: What "Thinking" and "Effort" Actually Mean

Wed, 15 Jul 2026 00:00:00 +0000

You may have noticed newer AI models pause and “think” before they answer, sometimes showing a summary of their working-out. These are often called reasoning models, and the pause is a real, deliberate feature, not a loading spinner. Understanding what it does explains why these models are better at hard problems, and why they can be slower and more expensive.

The old way: answer immediately #

A standard language model predicts its answer one chunk at a time and starts producing the final response right away. For simple questions this is perfect. But for a problem that needs several steps of reasoning (a math word problem, a tricky bit of logic, a multi-part coding task) answering immediately is like being forced to blurt out the answer to a hard question with no time to think. Sometimes you get it right, often you do not.

What Is MCP (Model Context Protocol)?

Wed, 15 Jul 2026 00:00:00 +0000

Once you understand that AI agents work by using tools, an obvious problem appears. Every useful integration (connecting an AI to your files, your GitHub, your calendar, your database) has to be built. And historically, every AI app built every one of those connections itself, in its own way. MCP, the Model Context Protocol, is the standard that fixes that.

The problem: everyone reinvents the same plug #

Say you want AI tools to be able to read your company’s wiki. Without a standard, the situation is a mess:

What Is RAG (Retrieval-Augmented Generation)?

Wed, 15 Jul 2026 00:00:00 +0000

A language model only knows what it absorbed during training, frozen at some point in the past. It has never seen your company’s documents, your notes, or anything that happened after its training cutoff. So how do AI tools answer questions about exactly those things? The most common answer is RAG, short for retrieval-augmented generation, and the idea is simpler than the name.

The problem RAG solves #

Ask a plain model “what is our refund policy?” and it has two options, both bad. It can admit it does not know, or it can make something up that sounds like a refund policy. It was never trained on your policy, so it genuinely cannot know.

Prompt Engineering: How to Actually Get Good Answers

Tue, 14 Jul 2026 00:00:00 +0000

“Prompt engineering” sounds like a specialized skill, but it is mostly common sense once you know how a language model works. Since a model just continues your text in the most plausible way, the way you ask shapes what you get. A vague prompt gives a vague answer. A clear, well-set-up prompt gives a sharp one. Here are the techniques that actually move the needle.

Be specific about what you want #

The most common mistake is being too vague and expecting the model to read your mind. It cannot. It fills ambiguity with the most generic plausible answer.

Running AI Models Locally: Open Weights and Quantization

Tue, 14 Jul 2026 00:00:00 +0000

Most people assume AI only lives in the cloud: you send your text to a company’s servers, they run a giant model, and they send an answer back (and bill you per token). But you can also run capable models directly on your own computer or phone. No cloud, no per-message cost, nothing leaving your device. Here is how that works and when it is worth doing.

Open-weight models #

The models you can run yourself are called open-weight models. The “weights” are the giant set of numbers that make up a trained model, the thing it learned during training. When a lab releases those weights openly, anyone can download them and run the model on their own hardware, instead of it living locked on a company’s servers.

Why LLMs Hallucinate, and How to Reduce It

Tue, 14 Jul 2026 00:00:00 +0000

The most important thing to know about AI is that it will sometimes state something false with total confidence. A made-up statistic, a citation to a paper that does not exist, an API function that was never real. This is called hallucination, and it is not a bug that will be patched away next month. It comes straight from how these models work, and understanding why is the key to working around it.

How Large Language Models Actually Work

Mon, 13 Jul 2026 00:00:00 +0000

Every AI chatbot, coding assistant, and agent you have used is built on a large language model, or LLM. They can feel like they understand you, reason with you, and know things. Underneath, they are doing something far simpler than that impression suggests, and understanding what it is explains most of their strange behavior.

The one-sentence version #

An LLM predicts the next chunk of text, given everything before it, over and over.

What Is a Token, and Why It Shows Up on Your Bill

Mon, 13 Jul 2026 00:00:00 +0000

If you spend any time around AI, you will keep bumping into the word “token.” The context window is measured in tokens. Pricing is per token. Bills go up when you use more of them. It is the basic unit everything in AI is counted in, and it is worth five minutes to understand.

What a token actually is #

A token is a chunk of text. Not quite a word, not quite a letter, but something in between. Before a language model can work with your text, it chops it into these chunks, because the model does not operate on raw letters or whole words, it operates on tokens.

Context Windows, and Why AI Agents Forget

Sun, 12 Jul 2026 00:00:00 +0000

If you have used an AI tool for a long session, you have probably felt it start to slip: it forgets something you said earlier, loses track of a decision, or repeats itself. That is not the model being dim. It is running into the context window, the single most important constraint to understand about how these tools behave.

What a context window is #

A language model can only consider so much text at one time. That maximum is its context window, and everything the model is working with has to fit inside it: your instructions, the entire conversation so far, any files or data it has been given, and the results of every tool it has used.

Free Things to Do in San Francisco That Are Actually Worth It

Sun, 12 Jul 2026 00:00:00 +0000

San Francisco is brutally expensive to live in and almost free to actually enjoy. The best things here (the coastline, the parks, the views) do not cost anything, and the city is dense enough that most of them are a walk or a cheap bus ride apart. You do not need to spend money to have a great day here. You need to know where to go.

The parks #

Golden Gate Park. Bigger than Central Park and stuffed with free things: the botanical garden has free hours for residents, the bison paddock is genuinely free bison, and there are lakes, windmills, and endless trails. You could spend a dozen full days here and not repeat.
The Presidio. A former army base turned enormous park, with forests, overlooks, and some of the best views of the Golden Gate Bridge in the city. Free, and criminally underused.
Lands End. Cliffside trail along the northwest corner with ocean on one side and the bridge in the distance. Walk out to the labyrinth and the ruins. This is one of the most beautiful free walks in any American city.
Twin Peaks. The classic panoramic view over the entire city and bay. Free, and best at sunset or after dark when the whole grid lights up.
Dolores Park. The city’s living room on a sunny day. Bring a blanket, watch the whole town show up, spend nothing.

The coast and the ruins #

Fort Point. The Civil War-era brick fort tucked directly under the south end of the Golden Gate Bridge. Free to walk through, and the view straight up at the bridge is unreal. One of the best free spots in the city and one most residents somehow never visit.
Sutro Baths. The ruins of a giant 1890s public bathhouse, right on the ocean next to Lands End. Free to wander, atmospheric as anything, especially in the fog.
Ocean Beach and Baker Beach. Miles of Pacific coastline. Ocean Beach for the long flat walk and the sunset, Baker Beach for the framed shot of the bridge. Both free, both spectacular.
Crissy Field. Flat waterfront path along the bay with the bridge dead ahead. Perfect for a walk, a run, or doing absolutely nothing on the grass.

Free museum days #

Most of the big San Francisco museums have free days, usually once a month, and several are free year-round for residents of certain counties or on specific days. The de Young, the Legion of Honor, the Academy of Sciences, and others all have free or discounted access if you time it right. Check each museum’s site for its free day and build your month around it. And your library card unlocks free museum passes too, on top of all the streaming.

Tool Use and Function Calling, Explained

Sat, 11 Jul 2026 00:00:00 +0000

Here is a puzzle. A language model does exactly one thing: it produces text. It cannot search the web, read a file, or query a database. And yet AI agents do all of those things constantly. How?

The answer is tool use (also called function calling), and it is the mechanism that underpins every agent. Once it clicks, the whole idea of agents stops seeming magical. The model still only produces text. The trick is what that text is allowed to say.

How Coding Agents Work Under the Hood

Fri, 10 Jul 2026 00:00:00 +0000

Coding agents (the tools that read your codebase, write changes, and run your tests) are the most mature kind of AI agent, and the most useful window into how agents work in general. Underneath the polish, a coding agent is just the agent loop pointed at a specific set of tools. Here is what those tools are and how the whole thing hangs together.

The tools a coding agent has #

An agent can only affect the world through its tools, so the tool set defines what a coding agent can do. The common ones are exactly what a developer uses:

How to Furnish an SF Apartment for Almost Nothing

Fri, 10 Jul 2026 00:00:00 +0000

San Francisco throws away astonishing furniture. This is a transient, expensive, high-turnover city full of people who are moving in a hurry, and a solid couch is heavy, so a huge amount of perfectly good stuff ends up free on the curb or given away rather than moved. I have furnished most of an apartment for close to nothing. Here is where it comes from and when to look.

The channels, ranked #

Buy Nothing groups. These are hyperlocal gift groups (on Facebook and the Buy Nothing app) where neighbors give things away for free, no strings. Join the one for your specific neighborhood. People post couches, desks, lamps, kitchenware, all of it, and it goes to whoever asks nicely. This is the single best source, because it is neighbors clearing out, not resellers.
Craigslist “free” section. The classic. Sort the SF Craigslist free category by newest and check it daily. The good stuff goes fast, so speed matters more than patience.
Facebook Marketplace, filtered to $0. Same idea, different crowd. Set the price filter to free and set an alert.
Nextdoor. Neighbors post free items and curb alerts here too. Worth a scan.
The actual curb. In this city, “free pile” is a genuine furnishing strategy. People leave good things on the sidewalk with a “free” sign, especially in denser neighborhoods.

Timing is everything: the end of the month #

Here is the insider move. Leases in San Francisco overwhelmingly turn over at the end of the month, so the last few days of any month, and the first day or two of the next, are a flood of people frantically getting rid of furniture they cannot move. The end of the month is your shopping season. Walk or bike a few dense residential blocks on the 30th, 31st, and 1st and you will pass real furniture on the sidewalk.

The Agent Loop: How AI Agents Actually Work

Thu, 09 Jul 2026 00:00:00 +0000

If you understand one thing about how AI agents work, make it this: the agent loop. It is the single mechanism that turns a language model from a thing that answers questions into a thing that gets work done. Once you see the loop, agents stop being mysterious.

The loop in one paragraph #

An agent runs in a cycle. The model looks at the current situation and the tools available to it. It either produces a final answer, in which case the loop ends, or it decides to use a tool. If it calls a tool, your program runs that tool for real, captures the result, and hands the result back to the model. The model looks at the new situation, including what the tool just returned, and decides again. This repeats until the model has nothing left to do and gives its final answer.

Cut Your Own Hair: A $25 Clipper Pays for Itself in One Haircut

Wed, 08 Jul 2026 00:00:00 +0000

A haircut in San Francisco runs anywhere from $40 to $70 once you tip, and you need one every three or four weeks if you keep it short. Do that math over a year and you are spending five to eight hundred dollars to have someone run a clipper over your head for fifteen minutes. I bought a clipper off Amazon for about $25 and it paid for itself before the first haircut was even finished.

What Is an AI Agent?

Wed, 08 Jul 2026 00:00:00 +0000

“AI agent” is one of those phrases that gets used constantly and explained rarely. Strip away the hype and it is a simple idea: an agent is a language model that has been given tools and a goal, and that works toward the goal in steps instead of answering in a single reply.

That one difference, acting in steps versus answering once, is the whole distinction. Everything else is detail.

How to Make Chinese Steamed Egg (Silky Water Egg)

Mon, 06 Jul 2026 00:00:00 +0000

Chinese steamed egg (zhengdan, sometimes called water egg) is a savory custard made from beaten eggs and water, steamed gently until it just sets into something silky and soft enough to eat with a spoon. It is delicate, soothing, and mild, the kind of dish Chinese families make for kids, for elders, and for anyone who wants something gentle and warm. Finished with a little soy sauce and sesame oil, it is quietly one of the best things you can do with a couple of eggs.

Cancel Netflix: Your Library Streams Movies for Free

Sun, 05 Jul 2026 00:00:00 +0000

Streaming subscriptions are the modern cable bill. Netflix, a music service, an audiobook app, a couple of the smaller ones, and suddenly you are bleeding sixty or seventy dollars a month to watch things you half pay attention to. Almost all of it is available for free, legally, from the public library. A San Francisco library card costs nothing and quietly replaces most of that stack.

The card is free and takes ten minutes #

An SFPL card is free to anyone who lives, works, goes to school, or owns property in San Francisco. You can sign up online in a few minutes and start using the digital services immediately, before you even pick up the physical card. That one card is the key to everything below.

How to Make Scallion Oil Noodles

Sun, 05 Jul 2026 00:00:00 +0000

Scallion oil noodles (cong you ban mian) are a Shanghai classic built on almost nothing: noodles tossed in fragrant scallion oil and a sweet, savory soy sauce, with the browned scallions scattered back on top. There is no meat, no vegetables to speak of, and yet the smell alone will pull people into the kitchen. It is proof that a couple of cheap ingredients, treated with a little patience, can beat almost anything you would order out.

San Francisco Rent Control, Explained by Someone Paying $1,680 for a $3,800 Apartment

Sun, 05 Jul 2026 00:00:00 +0000

My apartment rents for somewhere between $3,400 and $3,800 on the open market today. I pay $1,680. Same unit, same building, I just signed two years ago and never left. That gap is rent control, the single most valuable financial asset a normal person can hold in San Francisco, and most of the people who qualify for it do not understand it well enough to actually use it.

This is the long version, because it is worth understanding properly. Get this one thing right and it outperforms almost any raise, promotion, or side hustle you will ever chase in this city.

How to Eat for Free in San Francisco: the Tech Event Circuit

Sat, 04 Jul 2026 00:00:00 +0000

San Francisco runs on catered events. On any given weeknight there are dozens of tech meetups, product launches, hackathons, demo nights, and mixers across the city, and a huge share of them feed you, because free food is how they get bodies in the door. If you know where to look, you can eat well several nights a week without buying a single meal. Here is the circuit.

The calendar: Luma #

The single best tool is Luma (lu.ma). It is where the SF tech scene posts events. You can browse by night, filter for what is near you, and RSVP in two taps. Skim the descriptions for the magic words: “food and drinks provided,” “dinner served,” “catered,” “happy hour.” Those events are dinner. RSVP to a couple each week and you have a rotating meal plan that also happens to be full of interesting people.

How to Make Mapo Tofu

Sat, 04 Jul 2026 00:00:00 +0000

Mapo tofu is the famous Sichuan dish of soft tofu and ground pork simmered in a deep red, spicy, savory sauce, served over rice. It is bold where a lot of home cooking is gentle: hot, a little numbing from Sichuan peppercorn, and packed with the fermented, umami-rich flavor that makes Sichuan food what it is. A plate of it over rice is one of the most satisfying cheap meals there is.

How to Make Chinese Congee (Rice Porridge)

Fri, 03 Jul 2026 00:00:00 +0000

Congee (jook, or zhou) is rice cooked in a large amount of water until the grains break down into a soft, silky porridge. It is the comfort food of the Chinese kitchen: what you eat for breakfast, what you make when someone is sick, and what you fall back on when you want something warm, gentle, and almost free.

It is worth knowing because it costs pennies and asks almost nothing of you. A little rice and a lot of water, left to simmer, becomes a big pot of something soothing. It is endlessly adaptable through what you put on top, it stretches a tiny amount of rice into several servings, and it is about as easy on the stomach and the wallet as food gets.

The $2.05 Way to SFO: Take SamTrans, Not BART (BART Is for Tourists)

Fri, 03 Jul 2026 00:00:00 +0000

Everyone in San Francisco tells you to take BART to the airport. Everyone in San Francisco is wrong. BART to SFO is a tourist tax with a train attached. I take the SamTrans bus for $2.05, it drops me right at the terminal, and I keep the other eight-plus dollars. Here is the full breakdown, every option, honest about the tradeoffs.

Every way to SFO, side by side #

Option	Cost one way	Rough time from the city	Verdict
Uber / Lyft	$40 to $70	25 to 40 min	Unhinged for one person
BART	~$10 to $11	~30 min from downtown	Fast, and overpriced on purpose
SamTrans bus	$2.05	~45 to 60 min	The move

Same destination. One option costs five times more and lets you feel clever for “taking transit.”

Buy It Once: The Patagonia and Darn Tough Warranty Play

Thu, 02 Jul 2026 00:00:00 +0000

The most expensive way to buy clothes is cheaply. Fast fashion is a subscription you pay every few months when the seams blow out. The frugal move is the opposite: buy a small number of high-quality things, once, that are guaranteed for life, and then simply never buy them again. Two brands make this close to a cheat code.

Darn Tough socks: a real lifetime warranty #

Darn Tough makes merino wool socks in Vermont with an unconditional lifetime warranty. Not “lifetime against manufacturing defects.” Unconditional. You wear a hole in them, any hole, any reason, you send them back and get a new pair free, no receipt required. I own a handful of pairs. When one finally wears through, I warranty it. My sock budget is, functionally, zero for the rest of my life after the first buy-in.

How to Make Chinese Tomato and Egg Stir Fry

Thu, 02 Jul 2026 00:00:00 +0000

Tomato and egg stir fry (xihongshi chao jidan) is probably the single most-cooked home dish in China. It is soft scrambled eggs folded into cooked-down tomatoes, seasoned into a sweet and savory sauce, and spooned over rice. Almost every Chinese kid grew up eating it, and almost every Chinese adult can make it without thinking.

The reason it is a staple is simple: it is fast, cheap, comforting, and made from two things you always have. Ten minutes, two ingredients, one pan, and you have a complete meal that tastes like home. The trick that separates a good version from a flat one is a small pinch of sugar, which is the whole point of learning it properly.

How to Fund Part of Your Life With Hackathon Prizes

Wed, 01 Jul 2026 00:00:00 +0000

A hackathon is a weekend where companies feed you, sometimes house you, and hand out real cash to whoever builds something cool. If you can code even a little, it is one of the few side hustles that pays out in a single weekend, comes with free food the entire time, and makes you better at your craft while you do it. I have funded chunks of my life this way. Here is how it works.

How to Make Egg Fried Rice

Wed, 01 Jul 2026 00:00:00 +0000

Egg fried rice is the dish every Chinese kitchen falls back on. It is leftover rice, a couple of eggs, and some green onion, cooked fast over high heat until every grain is separate and toasty. It takes about ten minutes, costs almost nothing, and turns yesterday’s rice into a meal that tastes like you meant to make it.

What makes it worth mastering is that it is the ultimate use-it-up dish. A container of cold rice in the fridge plus whatever odds and ends you have becomes dinner. It is also the base recipe for a hundred variations: add a protein and it is a full meal, add vegetables and it is a balanced one. Learn the plain version first, because the technique is the whole game.

Patagonia's Warranty: What Actually Works (Not the Brochure Version)

Tue, 30 Jun 2026 00:00:00 +0000

Patagonia’s Ironclad Guarantee is the best in the outdoor business, but the brochure version, “we guarantee everything we make,” does not tell you how to actually use it to the fullest. Here is what works, what does not, and the specific moves that squeeze a decade out of one jacket.

The guarantee, briefly #

Two programs do the work. The Ironclad Guarantee covers actual defects in materials or workmanship with a repair, replacement, or refund. The Worn Wear program repairs normal wear and tear (the busted zippers, blown seams, and small tears that come from actually using the thing) on top of that. One covers “this was made wrong,” the other covers “I wore this hard for years.” Between them, there is very little that ends a garment’s life. Now the part that actually matters.

Stop Paying $40 a Month for AI: Use GLM and Qwen Instead

Tue, 30 Jun 2026 00:00:00 +0000

Somewhere along the way, paying $20 a month for ChatGPT Plus and another $20 for Claude Pro became the default. Forty dollars a month, five hundred a year, for tools most people use to write emails and reformat lists. For the vast majority of what you actually do with AI, you are massively overpaying. The frugal move is to default to a cheap model and only reach for the premium one when the task genuinely calls for it.

How I Host Everything for Free: The Best Free Tiers for Builders

Mon, 29 Jun 2026 00:00:00 +0000

People assume shipping a website or an app means paying for servers. It does not. The free tiers from the major platforms are now generous enough to run a real, live product on, with a real database and a real domain, for zero dollars a month. Every app and site I run, including this one, lives on free infrastructure. Here is the stack.

Hosting the front end: Vercel, Cloudflare, Netlify #

For websites and web apps, three platforms host you for free:

How to Beat Airline Bag Fees on Frontier and Other Budget Airlines

Mon, 29 Jun 2026 00:00:00 +0000

Budget airlines like Frontier and Spirit sell you a $29 ticket and then spend the rest of the transaction trying to claw it back in fees. The biggest one is the bag fee, and a single bag can cost more than your flight. Here is how to fly them without paying it, up to and including the nuclear option I demonstrate in the video below: wearing my entire wardrobe onto the plane.

How to Make Chinese Dumplings the Easy Way

Sun, 28 Jun 2026 00:00:00 +0000

Dumplings (jiaozi) are little parcels of seasoned meat and vegetables wrapped in a thin dough skin, then boiled, pan-fried, or steamed. They are one of the most traditional Chinese dishes there is, and one of the cheapest to make at home. The filling is savory and juicy, the wrapper is soft with a little chew, and a plate of them fresh out of the pot is hard to beat.

They also happen to be the perfect group food. Folding dumplings is something people do together: a cold day when everyone is inside, a date you want to impress, or just a night where a few friends sit around a table and make food and talk. The whole batch runs about $15 and feeds around six people (roughly three meals for two), and leftovers freeze beautifully, so it doubles as meal-prep. Here is the easy way, no homemade wrappers, no fancy technique.

How to Make Chicken Noodle Soup the Easy Way

Sat, 27 Jun 2026 00:00:00 +0000

Chicken noodle soup is the classic comfort bowl: tender chicken, soft noodles, and soft aromatics in a warm, savory broth. It is the thing you want on a cold day or when you are feeling under the weather, though honestly it is good any time. It is also one of the cheapest, most forgiving meals you can make, which is why it is a weeknight staple.

Part of the trick here is the pasta. It does double duty: it makes the soup filling and it lets you use less meat, which is easier on both your wallet and the planet. A big pot costs a few dollars, reheats great, and tastes even better the next day, so it is meal prep and comfort food at the same time. Here is the easy way, one pot.

How to Make Minestrone Soup the Easy Way

Fri, 26 Jun 2026 00:00:00 +0000

Minestrone is a hearty Italian vegetable soup built on a tomato base, loaded with beans and pasta and whatever vegetables you have around. It is thick, savory, and filling, the kind of bowl that eats like a full meal. It is also vegetarian, healthy, and, most importantly, cheap.

The real magic of minestrone is that it is the ultimate clean-out-the-fridge meal. The base is a handful of pantry staples, and after that you throw in any vegetables that need using up. That makes it both frugal and flexible: nothing goes to waste, and no two pots are ever quite the same. It also scales up easily and freezes well, so it is great for feeding a crowd or stocking lunches for the week. Here is the easiest version, one pot, no skill required. Simple Chinese guys only have one knife and use it for everything, and that is all you need here too.

The Easiest Chinese Soup: Pork and Seaweed, Two Ingredients

Thu, 25 Jun 2026 00:00:00 +0000

This is a classic Chinese home soup stripped down to its bare essentials: ground pork and dried seaweed in a light, clear broth. The seaweed gives it a gentle savory, slightly briny flavor, and the pork makes it just substantial enough to feel like a meal. It is subtle rather than bold, the kind of quiet, warming soup Chinese families make on a weeknight without thinking about it.

The reason to know this one is that it might be the easiest Chinese soup you can make. It has two ingredients, it comes together in one pot with almost no prep, and it is nearly impossible to mess up. It is cheap, it is comforting, and it is perfect for a cold day when you want something warm inside without any effort. It also scales down easily to a single serving, so it is a great “I just want soup” solution for one.

How to Make Chinese Stir Fry Over Rice

Wed, 24 Jun 2026 00:00:00 +0000

Stir fry is protein and vegetables cooked fast over high heat and tossed in a simple savory sauce, served over rice to make it a full meal. It is delicious, cheap, healthy, and quick, which is why it ends up on my stove constantly. One pot, one plate, dinner done.

What makes stir fry worth learning is how adaptable it is. It is less a fixed recipe than a template: pick a protein, pick some vegetables, flavor it simply, put it over rice. That makes it perfect for using up whatever is in the fridge and for eating well on very little money. It is also genuinely healthy, lots of vegetables, lean protein, and only as much oil and sauce as you add. Here is the simple way, plus which vegetables actually work and which ones do not.

How I Built a Private, On-Device Speech-to-Text App with Whisper

Sun, 21 Jun 2026 00:00:00 +0000

Most transcription apps upload your audio to a server to process it. If you are recording anything sensitive, a doctor’s visit, an interview, a private voice note, a business call, that is a hard no for me. So I built Private Transcribe: speech-to-text powered by Whisper, running entirely on your device. The audio never leaves your phone. Here is how it came together, and why one promise shaped every technical decision along the way.

Building Capybara Crossing: What Shipping a Casual Arcade Game Taught Me

Thu, 18 Jun 2026 00:00:00 +0000

After a long run of utility apps, I wanted to make something that was just fun. Something my niece could pick up in two seconds with no tutorial and no explanation. The “hop across endless lanes of traffic” formula is a classic for a reason, so I dropped a capybara into it and built Capybara Crossing: hop across roads, train tracks, and rivers, grab coins, try not to die. It turned out that a casual game is a completely different discipline from a tool, and it taught me things the tools never did.

How I Built an AI Chat App That Runs Entirely On Your Phone

Sun, 14 Jun 2026 00:00:00 +0000

Every mainstream AI chat app ships your conversations off to someone’s server. I wanted the opposite: a model that runs on your phone, messages that never leave the device, no login, no subscription. That app became Personal LLM, and getting a language model to behave itself on a phone was the most interesting engineering problem I have taken on alone.

The itch #

Three things kept bugging me about the mainstream AI apps I was using.