Portrait of Lê Xuân Lộc

Lê Xuân Lộc

Product designer & builder, Hanoi

I design and build apps to make life better.

I’ve made five iOS apps on my own, start to finish, each with a full payment flow. I’m looking for a UI or product design role.

Before the apps

In fourth grade I drew games in my notebooks, and my classmates paid to play them. In 2017, before university, I started building websites. At university the break between classes was only ten minutes, too short to queue for food. So I made an app for ordering at the cafeteria, and students didn’t have to stand in line and wait. Later I built web apps in R while I studied economics. I have always instinctively cared about how an interface makes people feel.

I co-founded Nala, a retreat near Hanoi, and since 2022 I have run its operations and its brand, with its staff and its brand partners. Online I became “Tớ là Lộc” on TikTok, with about 100,000 followers and videos that reached millions of views. For two years I was also content advisor to Nàng Mơ, a creator brand on TikTok, Facebook and YouTube. By 2024 the media work paid well, and it was taking me away from what I actually like doing.

So I stepped back from it and stayed home. People thought I was throwing my future away. I was building a new one: the apps I wished existed. Here are the five.

baseline On the App Store

Let the app do the thinking.

baseline home screen: a strip for the board, a year of activity, a prompt to choose 1 to 7 sessions each week, then TODAY with two exercises and a large orange START WORKOUT button.
baseline set-logging screen: 45 kg × 8 in very large type, SET 01 /03, a line naming the weight and reps gestures, and a full-width orange LOG SET button.
baseline set-logging screen.
The top of baseline’s home screen: a BOARD strip reading ONE SESSION TO JOIN, a year of activity as a grid, and a card that reads SET YOUR BASELINE. Choose 1 to 7 sessions each week. Train on any day.

  1. At the gym

    When I train I want physical exercise, not mental exercise.

    I get to the gym, open the app, and check today’s exercises. The thinking is the app’s job.

  2. First try: two drags

    I chose dragging early. Number keys mean tapping, and arrows eat the screen.

    In the first version there were two drags: one for weight, one for reps. I used it every day for nearly two weeks. It was fairly OK.

  3. The graph

    Using it, I felt something was missing: a graph that sets today’s weight against last time’s. I built it.

    Once the graph was there, one drag made more sense than two: up and down for weight, left and right for reps.

  4. Your turn

    Put a finger on the numbers and move. Up is heavier. Right is one more rep. The block is your set, drawn against last time.

  5. True to yourself

    At first it was only for me. I didn’t care if anyone else used it. Then I saw that might be why I wasn’t consistent, and staying consistent is the point of the app.

    So I added a social side, to feel some pressure. Not too much. I designed the board to score one thing: how true you are to yourself. How much you lift, and how often, don’t count. Say four sessions a week and do four, and you stand level with someone who says seven and does seven.

    Read the baseline essay baseline on the App Store

Lockeat On the App Store

Start with a photo.

From the age of 14 to 26, my weight went up and down. I worked hard and changed my body more than once. Then everything reversed, and each cycle left me with a little more shame.

Looking back over those twelve years, I realised I wasn’t broken. I had been trying to solve a long-term problem with periods of effort too intense to sustain. The healthy people I admired weren’t winning an enormous battle every morning. Eating well had become an ordinary part of their lives.

So I built Lockeat, a daily calorie tracker, around one principle: as little time in the app as possible, for as much effect as possible. You log a meal by taking a photo of it, and the estimate comes back in a second or two. The aim I set was a single touch. The app as shipped takes three: open the camera, take the photo, save.

I wanted discipline without the feeling of being disciplined. Lockeat doesn’t nag you to hit a daily target or make you obsess over every small thing you eat.

I prototyped the capture flow in Claude Design first. The app opens on a calendar: each day is a plain total, and one camera button sits at the bottom. There is no calorie ring and no progress bar on it. You log first and judge later.

Every figure is an estimate, not medical advice.

Lockeat on the App Store

Lockeat review screen over a photo of beef phở: 660 calories, 38 protein, 28 fat, 62 carbs, and the dish named in plain words above a Save button.
Lockeat day view: the day’s total, 2,399, at the top, joined by lines to the meals below. Two are in view, 763 at 08:00 and 1,365 at 12:00; a third line runs on past the crop.

Elsewhere On the App Store

Kept on your phone, quietly.

Elsewhere keeps personal media on the device, behind a discreet front.

You choose its Home Screen icon from twelve. iOS lets an app change its icon but not its name, so the name stays Elsewhere.

Elsewhere on the App Store

Elsewhere setting titled Home Screen icon: twelve icons to choose from, and a note saying the icon changes across the device while the name stays Elsewhere.
Elsewhere library: a three-column grid of sample photos of a fictional couple and keepsakes.

lifelong On the App Store

One person.

lifelong is a private space for exactly two people. No contact list, no feed.

Its assistant, Tơ, runs on the phone. It reads the shared chat only when asked, and only after both people have switched it on. Either of you can switch it off again.

lifelong welcome screen: One person. For life. A private space for exactly two people.
lifelong consent sheet, titled ‘Tơ & AI — private to the two of you’. Two notes say the assistant runs on the phone and that reading the shared chat needs both people to agree. Below, one row per person: ‘Me’ has a switch; ‘You’ is marked ‘hasn’t agreed’.

It is also the app I nearly abandoned. After a breakup I lost hope in it, and planned changes so large they were close to a rewrite.

I decided that one painful experience shouldn’t change the app’s picture of a healthy, happy relationship, so I kept the original. It is at about 30% of what I want it to be, and I’m still improving it.

lifelong on the App Store

Knowless On the App Store

One screen and one button.

Knowless live screen: translated lines with their originals beneath.

  1. Lost in Hong Kong

    When I was small I got lost in Hong Kong. I didn’t speak the language.

    On later trips, to China and to other countries where I don’t know the language, I couldn’t use mobile data. Each time I wished for the same thing: a translator that needs no connection, so I could understand people and talk with them.

  2. First try: a radar

    The first version was a radar that listened all the time. I imagined it could tell where a speaker was from sound alone.

    It couldn’t. My phone gives me no spatial audio from a recording, so nothing in it says where a voice is coming from. I read that a newer model might. I don’t own one. I stopped.

  3. Second try: eyes

    So I tried eyes instead of ears: both cameras at once, to find the face that was speaking.

    The phone ran hot, the battery drained, it froze. I dropped that too.

  4. My own ears

    On paper, noise reduction for loud places sounded reasonable. Then I listened to a noisy recording with my own ears. I couldn’t make out my own words.

    No amount of denoising would fix that, so I dropped the idea. That is what I mean by a human in the loop: I have to keep checking that what I’m asking for makes sense.

  5. What was left

    One screen and one button. Press, talk, translate.

    The first prototype also kept a conversation history. I cut that too, to ship in time for Shipaton, and still nearly missed the deadline. It is probably the simplest app I have made.

  6. Two days on one effect

    A very small feature turned out to be extremely complex: how the words appear. Speech becomes a guess, shown at once so it feels live. Then the guess is corrected. After a long sentence it is corrected again, with context.

    Each of those states has to look right, and so does the waiting in between. I went back and forth between the app and a design mock for more than two days.

    Some things shouldn’t be there. Not everyone notices.

    Read the Knowless essay Knowless on the App Store

Like Lego

When I was building several apps at the same time, things started to connect. A piece made for one fits another, like Lego. baseline began as the fitness part of lifelong; in August 2026 I split it out as its own app, and a few days later Lockeat came out of baseline.

The feed in baseline came from yet another app. I built that one, and only once it was running did I see the idea was too far from reality to work. Some days I regret the time it cost. But its feed moved into baseline and fit.

One set of rules

baseline, Lockeat and Knowless draw on a small design system I keep, called Ground & Signal. Its idea is that the apps look nothing alike and behave the same way. Its rules:

  • Two screens, then sheets.
  • One accent colour, for the item you are on and the one action worth taking.
  • Motion reports what is happening; it doesn’t perform.
  • The small type says the true thing.
  • No alerts or confirmations where an undo will do.
  • Report, don’t cheer: no streaks, no badges, no praise.

How I work

I prototype before I build, and I move fast. Lockeat became its own app on 26 August 2026. By 28 August its capture flow (an estimate card, a portion wheel, three swipe gestures) was a clickable prototype in Claude Design, and an AI-assisted review had gone through it screen by screen against what iOS allows. The next day the build spec was written from it, down to keeping the feel of the prototype’s motion.

At that stage I made two calls. An estimate is not a measurement, but I decided the app would still show one plain number, with the reasoning behind it streaming live. And one mock-up felt wrong to me: the cause turned out to be AI-generated photos among its sample food pictures. Since then Lockeat has had a rule for its pictures: real meals only, no AI-generated or studio stock photos.

I build with AI tools (Claude Code, Bitrig), using the best model I can get, and I stay closely in the loop. AI gives me many ideas that aren’t feasible, though sometimes it impresses me a lot. When I’m worn out I have the models review a plan first, then I read it for anything illogical. The decisions stay with me.

All five, in 36 seconds

On my phone the five sit in one folder, named Quality life.

Five apps for a quality life.

About

I try to make design one of the most important parts of everything I build, and I take each app from the first prototype to the App Store myself.

I’d like to bring that to a team. I’m based in Hanoi and open to UI and product design roles in Hanoi, in Ho Chi Minh City, or remote. I work from a fully equipped home office, on Vietnam time (GMT+7).

Contact

24m2space@gmail.com

Email me LinkedIn CV App Store

Reading this site

There are no captions under the pictures, so here is what they are. The phone screens are captures and compositions of the real interfaces, filled with sample content; the photos in Elsewhere’s library are generated and fictional, and one Lockeat day view is cropped. The clips are cut from my own demo and preview videos, with scripted sentences in Knowless. The three line drawings in the Knowless chapter (a radar, two cameras, a noisy waveform) are sketches made for this page, not captures. The two screens you can try are small recreations built for this page, not the apps. Two lifelong captures had iOS overlays retouched out (the AssistiveTouch button and the recording indicator), and the chapter numbers in the 36-second cut are masked; the apps’ own interfaces are as captured. Only the 36-second cut has sound, and only when you play it. It is demo footage, not a record of what each current release contains.

Photo credits

The Wikimedia Commons food photos used in Lockeat’s screens and video are listed here, resized and cropped; each stays under its own licence. The list covers the whole set, so not every photo appears on this page.

Licence texts: CC BY-SA 3.0, CC BY-SA 4.0, CC0 1.0. Headings are set in Fraunces (SIL Open Font License). Apple, iPhone and App Store are trademarks of Apple Inc.; this site is not affiliated with or endorsed by Apple.