Software engineer · systems and research

I build software that makes complex systems observable, reliable, and useful.

I work across runtime verification, robotics, embedded systems, and developer tooling—usually where performance, correctness, and practical constraints meet.

Explore selected work Download CV

01

Research

Two publications

Runtime verification and social robotics

02

Systems

1st of 30+ teams

Prioritized multi-threaded TCP server

03

Experience

TaleWorlds + Databin

Developer tooling and systems automation

02 / Selected work

Work at the boundary of theory and things that have to run.

Two recent projects where the implementation details are inseparable from the research question.

Three temporal-logic properties compiling into a shared subformula graph and a double-buffered linear runtime. — Shared formulas compile into one execution graph, then update together in a linear runtime pass.

01 Runtime verification 2026

LoomRV: Multi-Property Temporal Logic Monitoring

Compiled shared subformulas into one cache-conscious runtime instead of evaluating every property independently.

A high-performance C++ runtime verification framework for evaluating many temporal logic properties together through shared subformula execution.

C++20 · CMake · Docker · Runtime verification · Metric Temporal Logic

Read the case study

02 Robotics research 2025

Learning Social Navigation in Mobile Robots

Published at IEEE SIU 2025 after reducing social-norm violations against simpler navigation baselines.

A robotics research project on learning socially compliant navigation behaviors in dynamic pedestrian environments.

Python · PyTorch · Unity · Graph Attention Networks · Conditional Neural Processes

Read the case study

Browse the full project index

03 / Research

Published work, kept close to the implementation.

Google Scholar

01

2026 · arXiv preprint arXiv:2605.13668

Multi-Property Temporal Logic Monitoring

Arınç Demir, Doğan Ulus

A high-performance framework for monitoring multiple past-time LTL and MTL properties together by compiling shared subformulas into a common execution graph with a data-oriented, double-buffered runtime.

arXiv ↗Artifact ↗

02

2025 · 2025 33rd Signal Processing and Communications Applications Conference (SIU)

Mobil Robotlarda Sosyal Navigasyon Öğrenme / Learning Social Navigation in Mobile Robots

Berke Kartal, Arınç Demir, Erol Uğur, Y. Yıldırım, Zeynep Aktaş

A robotics research project on learning socially compliant navigation behaviors in dynamic pedestrian environments using simulation, graph-based interaction modeling, and trajectory prediction.

IEEE ↗Research code ↗

04 / Field notes