Andrey Leshchenko
Web developer | Bots, automation & small web fixes
New to Contra
Andrey is building their profile!
AURELIA: Next.js jewelry store with an owner admin panel
My own project. A demo jewelry store in Ukrainian: from the storefront to the owner's back office. The cover is a real screenshot of the admin panel on working data.
What works:
Catalog with product variants: size, plating, separate price and stock for each variant.
Cart and checkout with an AUR-XXXX order code. The total is calculated on the server, so it cannot be swapped from the browser.
Customer account: order history, favourites, saved searches.
Promo codes: percentage or fixed amount, usage limit, and the total never goes negative.
Admin: catalog, stock with a movement log, orders, customers, analytics, audit log.
sitemap.xml, robots.txt, Product and Organization markup.
Stack: Next.js 15, React 19, TypeScript, Prisma, PostgreSQL.
Honest limits: payment and delivery here are manual. There is no real acquiring, no carrier API and no fiscalisation — those are the shop owner's separate decisions, not forgotten details.
1
11
Nero: Discord publishing bot with PNG card rendering
My own project. A bot for a community that keeps a chronicle of an invented world. Every piece goes draft → queue → scheduled publication. The cover shows real PNG cards the bot generated, not a design sample.
What works:
Card rendering to PNG: background, frame, caption, text wrapping that adapts to length.
Scheduler: an item goes into the queue and is published at the given time.
Send queue with retries, so a publication is not lost to a network failure.
Audit log: who called which command and what changed.
Permission levels: member, editor, administrator.
Export and backups, with an integrity check before restore.
Stack: Python, discord.py (http://discord.py), SQLite, Pillow.
Tests: about 40 pytest files; scheduler, queue, audit and restore-from-backup are covered separately.
1
10
Health Control: offline health tracker in vanilla JS
My own project. A single-page app with no backend: everything is calculated in the browser and the data stays on the computer. The cover shows six real tabs. The records on them are demo data — I do not show my own.
What works:
Daily records: food, training, sleep, wellbeing, body measurements.
Analytics: calories, protein and fibre, weekly totals, best and worst day.
A fill-quality score. "Don't remember" counts as an honest gap and lowers the day's score instead of being disguised as a filled field.
An RPG sheet: level, index, body map, progressions — so that keeping a diary is not boring.
Trend charts for weight, measurements and activity.
JSON export and import to move between browsers.
Stack: HTML, CSS, JavaScript. No frameworks, no build step.
Tracker data is never sent to a server: the working state lives in the browser's localStorage. The page does load external web fonts, so that claim is about user data, not about the total number of network requests.
1
11
MacroLens AI: Flutter food logging app
My own project. The Flutter base of an app that estimates the calories in a meal and keeps a food diary. The cover shows three real screens from a built release: the daily total, a meal recognition result, and manual entry.
What works:
Daily total: calories, protein, carbs and fat against the goals in the profile.
Meal recognition with a confidence score and a from–to range instead of a single number.
Editing items: change the weight, remove an item, add your own — the total recalculates.
Manual entry for when recognition fails.
Meal history by day.
A free recognition limit and a paid subscription screen.
Local data deletion with a confirmation step.
Stack: Flutter, Dart. One codebase for iOS and Android.
Data stays on the device. The app shows an estimate, not a medical recommendation — and says so in the interface.
0
9