Software, data and AI engineering

Mete Serpil

I write backend services in Spring Boot and Node.js, build real-time data pipelines, and ship AI systems that test themselves.

Mete Serpil
  • Java
  • Spring Boot
  • REST
  • Node.js
  • Hibernate
  • JPA
  • Python
  • PyTorch
  • BERT
  • nnU-Net
  • LLM orchestration
  • Apache Spark
  • MongoDB
  • SQL
  • C++
  • ReactJS
  • Cloudflare Workers
  • Docker
  • CI/CD
  • Git
  • Linux
216score on the Freya hackathon test set
5months building backend services at Sebit
4projects across AI agents, VR, data and NLP
1IEEE paper on technology in neurosurgery

About

Computer Engineering graduate from Çankaya University, 2026. I work across full-stack web development, real-time data processing and AI-driven applications: five months of backend work at Sebit, a summer on the frontend at İŞ Yazılım, and research that became an IEEE-published paper on VR and medical imaging. I like problems where the hard part is making a system behave the same way twice.

Based in
Ankara, Türkiye
Focus
Backend, data engineering, AI
Languages
Turkish (native), English (C1), German (C1)
Nationality
Turkish, German

Experience

Two internships, 2024 to 2026.

  1. Sebit Eğitim Bilgileri ve Teknolojileri A.Ş.

    16 Jan 2026 – 22 May 2026

    Backend Developer Intern · Ankara, Türkiye

    • Developed and maintained scalable RESTful APIs in Spring Boot and Node.js for Sebit's educational technology platform.
    • Collaborated with frontend and product teams to design reliable backend services, ensuring high availability and performance across multiple applications.
    • Spring Boot
    • Node.js
    • REST
  2. İŞ Yazılım

    Jul 2024 – Aug 2024

    Frontend Developer Intern · Ankara, Türkiye

    • Built and optimized websites with ReactJS, creating dynamic user interfaces and extending existing functionality.
    • ReactJS

Projects

Four projects and one publication.

Hackathon · Oct 2026

Build the Builder

An AI agent builder for Turkish banking voice calls, built at the Freya hackathon.

  • Python
  • LLM orchestration
  • Cloudflare Workers
  • REST
See it on GitHub ↗
  • Built a Python tool that reads a bank's unstructured documents — briefs, emails, meeting notes, API contracts, PDFs — and generates a working Turkish phone agent, with no manual edits.
  • Designed a hybrid setup: the LLM handles conversation while deterministic code handles date maths, offer rules and closing logic, which made behaviour repeatable and cut response time.
  • Built a self-testing loop: the tool writes its own test calls, runs them against simulated customers, scores the results and repairs the agent.
  • Tested against 250+ edge-case scenarios, reaching 216/250 on an independent test set and 9/10 on the public benchmark.
  • Built KVKK safeguards into the generated agents — data minimisation, no exposure of card numbers — and tested them against adversarial prompt-injection scenarios.
  • Deployed a mock banking API on Cloudflare Workers covering seven services with contract-exact responses and error codes.

VR · Medical imaging

MedVisVR

Unity, C#, Python, nnU-Net

  • Built interactive visualization features in Unity (C#) for a VR medical imaging platform on Meta Quest 3.
  • Integrated nnU-Net segmentation pipelines in Python to convert MRI and CT scans into 3D anatomical models.
  • Optimized real-time rendering and worked on VR interaction, UI and medical image processing workflows.
  • Unity
  • C#
  • nnU-Net
  • Meta Quest 3
See it on GitHub ↗

Data engineering

Earthquake Detection and Report System

Apache Spark (Scala), Python, MongoDB

  • Designed a real-time disaster response system using Spark, Python and MongoDB for distributed processing, ETL and Twitter API ingestion.
  • Built pipelines that identify emergency gathering areas and incident hotspots from social media streams as they arrive.
  • Spark
  • Scala
  • MongoDB
  • ETL
See it on GitHub ↗

NLP

Offensive Language Detection in Turkish

Python, PyTorch, BERT

  • Fine-tuned BERT transfer models to detect and classify toxic language in Turkish text.
  • Compared NLP preprocessing techniques to improve accuracy while preserving the context of informal Turkish.
  • PyTorch
  • BERT
  • Transformers
See it on GitHub ↗
IEEE · IISEC 2026

Transformative Technologies in Neurosurgery: a Systematic Literature Review

M. Serpil, A. B. Çetinkaya, M. Y. Özcan and S. Ataş. 2026 5th International Informatics and Software Engineering Conference (IISEC), pp. 622–627. DOI: 10.1109/IISEC69317.2026.11418476 ↗

Database

The same CV as relational tables. Click a table to query it, then a row to expand the record.

psql — mete@cv

Skills

Languages

  • Python
  • Java
  • C++
  • C
  • SQL

Backend and data

  • Spring Framework
  • Spring Boot
  • REST
  • Hibernate
  • JPA
  • JDBC
  • JAX-RS
  • Maven
  • Gradle
  • Node.js
  • MongoDB
  • Apache Spark

ML and AI

  • PyTorch
  • BERT / Transformers
  • nnU-Net
  • LLM orchestration

Frontend

  • ReactJS

DevOps and tools

  • Linux
  • Git
  • GitLab
  • Docker
  • CI/CD
  • Agile Scrum

Education

2022 – 2026

Çankaya University

Bachelor of Science in Computer Engineering · Ankara, Türkiye

TurkishMother tongue
EnglishC1 across all four skills
GermanC1 across all four skills

Contact

I'm looking for my first full-time engineering role.

Backend, data engineering or AI.

meteserpil06@gmail.com