Delen
Bewaren

Kavian

Machine Learning Engineer
nog geen reviews Leeuwarden

Over deze freelancer

With 5+ years of end-to-end ML experience and an MSc in Computer Vision & Data Science (NHL Stenden), I've shipped real-time inference systems, forecasting pipelines, and computer vision models used by 50,000+ users, cutting latency by up to 200ms along the way. My focus: deep learning, computer vision, and time-series forecasting. My stack: Python, PyTorch, MLflow, Docker, FastAPI, Flask, covering the full lifecycle from raw data to deployed, monitored model. What sets me apart is the combination of research depth and production mindset. I care about model performance, but also about inference logging, data drift, and the reliability of what I ship. Currently based in the Netherlands and open to ML Engineer roles where I can build scalable systems and solve real business problems.


Opleiding


Werk & Ervaring

M
20-06-2019 — 20-06-2025
Machine learning engineer
TOSAN

• Designed and delivered machine learning systems for an online gas consumption forecasting platform serving over 50,000 users, packaging models for deployment using Docker. • Built and maintained end-to-end ML pipelines, including data preprocessing, database integration, model training, evaluation, deployment packaging, and monitoring using Python and MLflow. • Owned the full machine learning lifecycle, covering feature preparation, relational database integration, API development, model training, evaluation, and production readiness. • Led the development and improvement of time-series forecasting models for the MAHAN project, significantly contributing to model performance, reliability, and real-time usability. • Improved system latency by approximately 200 ms through feature optimization and multiprocessing techniques, enhancing real-time inference performance. • Developed REST APIs using Flask and FastAPI to serve models and enable integration with production systems. • Implemented logging and monitoring pipelines for model performance, inference outputs, and data drift to support production stability and continuous improvement. • Contributed to CI/CD workflows by preparing containerized ML services and integrating code into deployment pipelines using Bitbucket and GitLab.


Certificeringen


Portfolio


Reviews

nog geen reviews
5 Sterren
0%
4 Sterren
0%
3 Sterren
0%
2 Sterren
0%
1 Sterren
0%

€ 35 / uur
  • Locatie Leeuwarden
  • Categorie Development & IT
    Techniek & Engineering
  • Geverifieerd ongeverifieerd
  • Lid Sinds 20-06-2026