Antonis Stellas - Personal Website

Antonis Stellas

Freelancer Data Scientist | EngD Data Science

About me

I am a Freelance Data Scientist | ML Engineer. I currently work at a start-up called Nanometrisis and undertake projects on Upwork. I hold an Engineering Doctorate in Data Science from the Technical University of Eindhoven, Netherlands. There, I had the opportunity to work on projects with multiple companies, such as Omron, ASML, TE Connectivity, and Brabant Water. I have a master's degree in Nanotechnology and a bachelor's degree in Applied Mathematics and Physics. During this period, you will find me active on DataTalksClub's

Slack, and on Hugging Face's

Discord, where I contribute to the community's vision course, as well as on Evidently AI's

Discord. I also love playing tennis and watching the NBA!

Projects

Personal Projects

Semiconductor quality prediction and pipeline

An MLOps pipeline on semiconductor quality prediction. The projects contains semiconductor sensor data and classifies the end product as Pass or Fail. There are ~580 sensor features that are used. Stack used: MLflow, Prefect, AWS S3, Docker, MongoDB, EvidentlyAI, Prometheus, Grafana

Tracking DataTalksClub's Youtube playlist stats

A batch and stream process for tracking Youtube's API stats. Stack used: Kafka, Prefect, dbt, GCP Storage, BigQuery, Terraform

Engineering Doctorate (EngD) Projects

Implementation of Omron's self-learning assembly line demo

Built system that uses a machine learning model as well as a collaboration of different user modes that receive user's input data and project optimal assembly step sequences.

Statistical process control for IPC assembly line's checker machines

I conducted root cause analysis for a quality checking machine issue by querying sensor data, analyzing it, and visualizing findings to identify malfunctioning parts. To prevent future problems, I implemented statistical process control measures for early detection on the assembly line.

Dust particle size estimation

To improve wafer production, ASML utilized spectroscopy image data to estimate the size of dust particles on their reticles. By creating and testing new features, I optimized the estimation process, leading to more accurate predictions using ML models.

Identification of opportunities for profitable growth in product sales

The analysis examined factors impacting profit margins, pricing strategies, and customer relationships, using PCA to reduce the dimensionality of the dataset and clustering techniques to identify distinct customer segments, each with specific actionable insights to enhance profitability.