Hasan Hüseyin
Yurdagül

Machine learning engineer. Consultant. Deep thinker.
I build things from first principles.

ML Engineer Consultant Arch + Hyprland Neovim → Helix 27 Turkey

/about

I'm a machine learning engineer working remotely from Turkey, and a consultant to R&D Centers on AI and ML. I hold a degree in Computer Engineering and I'm currently pursuing my Master's in Artificial Intelligence.

Cow — the real animal

/Articles

Thinking on engineering, math, systems, and life.

All entries

/projects

Case studies of systems I've built end-to-end.

MLStudio — From a 30-Company Desktop Tool to a Web ML Platform

Deployed

A deployed Streamlit platform for training, validating, packaging, and testing regression and neural time-series models, rebuilt from a proven Tkinter product.

Fishometry — Measuring Fish Without a Reference Object

Research complete · Thesis writing

Completed master's-thesis research on estimating fish length from a single image without a ruler or physical reference in the frame.

MdsAI — A Human-Reviewed LLM Platform for Support

Deployed · Active R&D

A deployed AI platform that turns four years of noisy helpdesk history into retrieved evidence, answer drafts, routing suggestions, and process intelligence.

Electricity Consumption Forecasting for the Turkish Grid

Done

A completed forecasting application that benchmarks zero-shot TimesFM forecasts against EPİAŞ's official hourly load-estimation plan.

Teus — Forecasting Flash Floods Before They Happen

Analytics complete · App in development

A hydrologist-validated deep learning system that forecasts river discharge up to 72 hours ahead using sensor history and external meteorologist-produced rainfall forecasts.

All projects

/tech

I've been a Linux user since 2016. The journey has been long — from Mint to Arch, a detour through NixOS (beautiful pain), a macOS era (M1 was magic), and finally back home to Arch + Hyprland. Freedom matters. Window managers matter.

distro_hop.sh
$ [2016] Linux Mint # the beginning
$ [2016] Ubuntu # the obvious next
$ [2016] Pop!_OS # the hyped children
$ [2016] Fedora # usable, but not fresh
$ [2016] Manjaro # intro to arch, but gentle
$ [2017] Arch Linux # the real deal
$ [2020] NixOS # beautiful pain
$ [2021] macOS (M1) # incredible hardware
$ [2022] Arch (again) # yearning freedom
$ [2022] macOS (M1) # disgusting software, but unbeatable hardware
$ [2025] Arch + Hyprland # returning home

Languages I enjoy

Python C Rust Zig Odin OCaml Lua JS/TS

I try every paradigm. Systems programming, functional, scripting — understanding the spectrum makes you a better engineer. I create things from scratch to understand them from first principles, it takes time, but it's worth it.

cs_amazement.log

The first thing that amazed me in CS was the linked list — this elegant, recursive data structure where every node points to the next. Then automata theory: the idea that a simple machine with states and transitions could model computation itself.

ai_amazement.log

In AI, autoencoders blew my mind — the idea that a network could learn to compress and reconstruct, discovering the essence of data. And GANs: two networks battling each other, creating something from nothing. Adversarial creation is pure genius.

editor_journey.md
~ neovim ── longtime user, broke every update
~ helix ── current editor, good defaults, stable and fine
~
~ # I value my time more as I age
~ # I don't have enough time to configure everything.
~ # I need things delivered, but I must be able to change them.
~ # Helix gave me the balance of freedom, and sanity

/work

Universal Software

Head of R&D AI
Jan 2024 - Present · Istanbul, Turkey (Hybrid)

Lead the AI R&D division directing municipal and enterprise AI strategy. Deployed LLM and sentence-similarity pipelines for support ticketing optimization that cut response times in half, and built automated categorization systems using sentiment embeddings. Also serving as the technical lead for international Horizon Europe consortium proposals such as FloodGuard and Pre-act.

Python LLMs Transformers Word Embeddings TF-IDF Clustering Horizon Europe Scoping

R&D AI/ML Consultancy

Independent Consultant
Sep 2020 - Present · Turkey (Remote)

Provide strategic machine learning, time-series forecasting, and R&D consultancy for flagship e-commerce and fintech platforms. Built semantic search and image classification systems for Inveon, developed end-to-end pricing and churn prediction pipelines on Azure for Moka, and shipped forecasting tools for Ikas and Robokobi.

Python FastAPI Streamlit Docker Kubernetes Azure Time-Series Forecasting Recommender Systems

M.Sc. Artificial Intelligence

Thesis Research
Aug 2023 - Feb 2026 · Çukurova University

Completed Master's in Artificial Intelligence with a thesis focused on monocular depth estimation and attention-driven vision networks. Developed models capable of estimating an object's physical length from a single photograph without requiring physical reference markers in the frame.

Computer Vision Depth Estimation Attention Models PyTorch Deep Learning
Full work history