Portfolio · Repository activity
12 Public repositories · 14 GitHub language categories · First-class BSc (Hons) Computer Science
Project facts checked 2 October 2026. Activity badges update from GitHub.
First-Class BSc (Hons) Computer Science · De Montfort University, Leicester, UK
Life, liberty, and the pursuit of reproducible applications.
I write backends and the web apps that run on top of them, and I have an unreasonable soft spot for the part most people skip: proving the thing actually works.
A benchmark from a single run tells you almost nothing — change the seed and it can tell you the opposite. So when I claim my optimiser beats a greedy baseline, that number comes from 120 scenarios × 30 seeds, reported as a mean with the spread next to it. A result without a distribution is just an anecdote with good lighting.
That habit came out of my degree and never left. Give me a dataset and I'll happily lose an evening to it.
A supply-chain matching platform with a FastAPI + PostgreSQL backend and a Next.js/TypeScript frontend. It ranks manufacturer–logistics offers by a customer-weighted cost, delivery-time and reliability score; a DEAP genetic search cross-checks that score against a greedy cheapest-first baseline.
The interesting part is the trade-off: faster and more reliable offers can cost more. The customer chooses the weights; the deterministic mode exhaustively ranks the available offers, so the genetic search can match that score but cannot improve its optimum. The recorded benchmark covers 3,600 evaluations (120 scenarios × 30 seeds) against the engine code:
| Metric | Greedy baseline | Optimised | Change |
|---|---|---|---|
| Composite fitness | 0.682 | 0.801 | +17.5% |
| Delivery time | 8.02 days | 4.67 days | 41.8% faster |
| Reliability | 0.824 | 0.891 | +8.1% |
| Raw cost | 21,296 KZT | 51,648 KZT | +142.5% — a deliberate trade |
These are recorded benchmark results, not a promise about every real supply chain. The seeded scenarios and exported measurements make the comparison inspectable.
The engine has a pytest regression suite; the application uses Argon2id, server-side sessions and rate limiting. Recorded benchmark data includes the scenario counts, seeds and measured trade-offs.
| Project | What it is | Demo / setup |
|---|---|---|
| intelli-factory | Weighted supply-chain matching — FastAPI, PostgreSQL, Next.js, DEAP cross-check | Visit |
| DrivePro_2 | Navigation-first Almaty carpooling PWA with route-based matching | Visit |
| portfolio | Interactive portfolio with a 3D project showcase | Visit |
| todo-webapp-refactored | PHP + MySQL task manager with shared lists and JWT authentication | Run locally |
| drivePro-website | Bilingual (RU/KK) site for an equipment-hire company | Visit |
| currency-exchange-bot | Button-only Telegram bot, API with a scraper fallback | @currenvy_bot |
| vue-folder-tree | Animated Vue 3 folder tree with selection, keyboard navigation and ARIA roles | Visit |
| react-starter-pro | React 19 + Vite starter with the lint/format/hooks/CI boring bits already done | Visit |
| qubly-landing | Responsive landing page. Plain HTML, CSS, jQuery | Visit |
| Project | What it demonstrates | Try it |
|---|---|---|
| student-course-hub | Server-rendered course catalogue and admin CMS with Deno, Oak and SQLite | Local setup |
| module-chooser-javafx | JavaFX desktop module selection, MVC and credit tracking | Desktop setup |
Snapshot: 2 October 2026. GitHub language bytes across public repositories; this shows repository composition, not proficiency or time spent.
Alongside those: DEAP for evolutionary algorithms, pandas and NumPy when I'm poking at data, and Recharts when a result needs to be looked at rather than read.
Looking for entry-level software and web development roles where I can build useful applications, learn from experienced engineers and test my work properly.
Certified: Meta Front-End Developer · Palo Alto Cybersecurity Foundation · Red Hat RH124.
If you've got a problem where the right approach isn't obvious and someone needs to measure which one actually wins — that's the job I want.
Reach me at igor_vuta@proton.me.
The snake eats my commits so I don't have to.