Dayanch Salarov Kemalovich — Backend Engineer
I am a fourth-year computational linguistics student at Oguzhan Engineering University, and a self-taught Django developer. My coursework is about how machines parse human language; my work is about making services that stay up.
Most of what I build is Python — REST APIs, text-processing pipelines, and NLP tooling — deployed behind Nginx with Redis and Docker. I work in Turkmen, Russian and English, which is useful when the data is multilingual.
A Turkmen language model built from scratch: text corpus, tokenizer, GPT pretraining on a rented GPU, chat training, and a web UI to talk to and compare the models.
43 to 71 out of 100 on a fixed test across three models (69 under stricter scoring)
Python, PyTorch, GPT, Tokenizers, Django
Production-ready Django REST API — AI chat (Gemini), i18n (en/ru/tk), Celery, soft-delete models, and Unfold admin.
207 automated tests; content served in three languages
Django, DRF, Celery, Redis, Gemini AI, Python
React + TypeScript site with a terminal-inspired design: typewriter animation, dark/light mode, AI chat widget, skeleton loading, per-field form validation, three languages and prerendered content for search engines.
a page per language that search engines can read without JavaScript
React, TypeScript, Vite, CSS
English / Russian / Turkmen API response localization via Django's LocaleMiddleware and gettext_lazy.
API content in English, Russian and Turkmen, falling back to English
Django, gettext, i18n, Python