Atoms to Algorithms

Ozkan Ceylan

I'm an AI and Robotics Engineer with a mechatronics background. I enjoy working on problems where machine learning meets real hardware. My current work focuses on learned manipulation — training imitation learning policies such as ACT and VLA for humanoid robots — while also building production AI systems that operate reliably outside research environments.

I started my career in industrial automation, working with manipulators and AGVs before moving into data science and machine learning. Since then I've worked on production forecasting for the energy sector and, more recently, helped build an autonomous last-mile delivery vehicle. Having worked across mechanical design, embedded systems, and machine learning, I'm used to solving problems that span the entire robotics stack.

Core Philosophy
“Real robots are noisy, imperfect, and unpredictable. Robust engineering means building systems that embrace this complexity rather than ignoring it.”
Professional Experience
FEB 2026 — PRESENT

AI Engineer

@Freelance
Berlin

Agentic AI systems and automation for clients (LangGraph, n8n, RAG pipelines). In parallel, ongoing research on bimanual humanoid manipulation — end-to-end ACT policies on the Unitree G1 in MuJoCo with a natural-language task dispatcher over ROS 2.

APR 2024 — FEB 2026

AI Engineer

@Area99
Berlin

Built an autonomous last-mile delivery vehicle from blank chassis to working prototype: embedded control and real-time actuation, ROS 2 perception and sensor fusion. In parallel, shipped enterprise RAG and LangGraph multi-agent systems on self-hosted Kubernetes with ArgoCD CI/CD.

MAY 2023 — JULY 2025

AI Engineer

@Greenbox Mobile Energy
Berlin

Production ML forecasting for the German energy market: EV charging station occupancy, day-ahead imbalance prices, and solar power production. Standardized the ML lifecycle with MLflow and containerized deployment.

OCT 2021 — OCT 2022

Project Engineer

@Dalmec Manipulators
Izmir

Industrial manipulator design for automotive, food, and logistics customers, with payloads from 50 kg to 1000 kg. Custom end-of-arm tooling for an automotive dashboard installation arm.

OCT 2020 — OCT 2021

Technical Purchasing Engineer

@Volt Electric Motors
Izmir

Robotic-cell hardware procurement, supplier qualification, and supply chain optimization.

AUG 2019 — JULY 2020

Project Engineer

@Bosch Thermotechnology
Manisa

AGV integration solutions for warehouse logistics automation.

2014 — 2020

Mechatronics Working Student

@Karaca Mekatronik
Izmir

Conveyor-based production line automation during undergraduate studies: PLC programming, sensor integration, and electromechanical assembly.

Technical Arsenal

Robotics & Simulation

ROS 2Micro-ROSMuJoCoPinocchioGazeboImitation Learning (ACT, VLA)Kinematics & ControlSensor FusionDomain Randomization

Embedded & Edge

STM32ESP32CAN BusFreeRTOSC++NVIDIA JetsonONNX / TensorRT

AI & Agentic Systems

PythonPyTorchTensorFlowLangGraphRAGLLMs / VLMsvLLMComputer VisionMLflowPhoenix (Arize)

Cloud & DevOps

KubernetesArgoCDTerraformDockerAWSFastAPICI/CDLinux
Academic Foundations
2022 — 2024
M.Sc. Data Science
University of Europe for Applied Sciences Berlin
2014 — 2020
B.Sc. Mechatronics Engineering
Izmir Katip Celebi University Izmir
Network

Open to interesting problems.

Always happy to discuss robotics, AI engineering, machine learning, autonomous systems, or opportunities to build impactful products.

Initialize Contact