Looking for a job

MERT YILDIZ

@mertyildiz

Machine learning engineer at BetterSea building maritime LLM extraction and trading systems; founded Seavec’s vessel-routing platform.

Germany

Message

At BetterSea, I designed and built a model-agnostic LLM pipeline that extracts structured data from maritime fuel and compliance certificates. I benchmarked 16 models and cut inference cost by 82% with no loss of extraction accuracy.

I also built a live EUA carbon-allowance trading system and implemented FuelEU Maritime, EU-ETS and UK-ETS calculation engines. I led API integrations for vessel reports across more than 10 data providers and built monitoring that surfaces silent feed failures.

At Seavec, which I founded and develop part-time, I built an AIS data pipeline and a maritime routing platform. Its route graph covers the North and Baltic Seas, and my Rust routing engine serves any-point-to-any-point queries in under 300ms.

At Fraunhofer CML, I researched and deployed maritime ML, including vessel trajectory forecasting, geomagnetic navigation and perception and autonomy for the SeaLion unmanned surface vehicle. Earlier, I developed soft robotic grippers and quadrupedal robots at Hamburg University of Technology and Bilkent University.

Experience

Work history, roles, and key accomplishments

Current

Senior Data & Machine Learning Engineer

Current

BetterSea

Mar 2026 - Present (8 months)

Designed and built a model-agnostic LLM document-extraction pipeline for maritime fuel and compliance certificates, cutting inference cost 82% with no loss of accuracy. Built a live EUA carbon-allowance trading system and implemented regulatory calculation engines for FuelEU Maritime, EU-ETS, and UK-ETS compliance.

LLM Documentation Extraction Harness STOMP WebSockets REST PostgreSQL Regulatory Compliance Data Pipelines Monitoring

Current

Founder & Sole Engineer

Current

Seavec

Sep 2025 - Present (1 year 2 months)

Built the AIS ingestion and storage pipeline behind the product, handling 776M vessel positions. Designed and implemented the core algorithm for unsupervised extraction of a directed maritime route graph, and wrote a Rust routing engine serving queries in under 300ms.

AIS Data Pipelines Rust Axum Tokio Svelte Nginx Martin Anomaly Detection Routing

Applied Data Scientist and ML Engineer

Fraunhofer CML

Sep 2021 - Sep 2025 (4 years)

Researched and deployed maritime ML across EU- and BMWE-funded projects, including vessel trajectory forecasting and reinforcement learning based anomaly detection. Delivered the perception and autonomy stack for the SeaLion unmanned surface vehicle, and drove projects through their full life cycle.

LSTM Transformers Autoencoder Reinforcement Learning YOLO V5 Path Planning Sensor Fusion DSPy Project Management

Research Assistant

Fraunhofer CML

Jan 2020 - Sep 2021 (1 year 8 months)

Delivered a low-cost IoT air-quality monitoring system for the EU-funded SCIPPER project, including LSTM regression models for multivariate air-quality index prediction.

IoT LSTM regression Air Quality Monitoring

Research Assistant

Hamburg University of Technology

Jan 2018 - Jan 2020 (2 years)

Developed the mathematical motion model and CAD designs for soft passive-gripper robotic arms.

Motion Modeling CAD Robotics

Research Assistant

Bilkent University

Jan 2015 - Jun 2017 (2 years 5 months)

Designed and manufactured soft quadrupedal robots, including gait control systems for motor-actuated actuators and electrically conductive silicone soft sensors.

Robotics Movement Control Soft Sensors