PORTFOLIO · STUDENT AT BREDA UNIVERSITY OF APPLIED SCIENCES
Certified · AI Engineer for Data Scientists
Associate · DataCamp
I'm a 3rd year data science and AI student with hands on experience in natural language processing, computer vision and with the MLOP's lifecycle. Throughout my school years I had been working on multiple interesting machine learning projects, this portfolio summarizes my key works and studies.
01 / SELECTED PROJECTS
An end-to-end MLOps system that turns video or audio into per-sentence emotion predictions. Built for a TV production client who wanted to understand which parts of episodes engage viewers emotionally.
An end-to-end computer vision and robotics pipeline for automated inoculation of plant root tips in a PyBullet simulation environment.
Real-Time audio transcription soundboard, a low-latency pipeline that captures live audio, transcribes it via streaming ASR, and triggers sound effects on regex keyword matches, engineered end-to-end for sub-second response from spoken word to playback.
Paperswap borrows the Tinder interaction model for reading news. Instead of scrolling an endless list, you get one new at a time, and each new has a short ML based summary on the back, if the user would not want to read the whole article just a bulletpoint summary.
Applied behavioural-cloning to teach an agent to play CounterStrike 2 and building a semi ML based framework to derive decisions based on probability mapping of enemies from the minimap.
Developed a pipeline to detect the most common skin-diseases (4 classes). The goal of the project was to build an AI driven pre-screening system, the project would have decreased the length of dermatologist appointments via providing AI driven insights to the doctor before the patient would arrive to the appointment.
AirBreda is an end-to-end cloud data pipeline built on AWS that collects hourly NO2 and traffic data for the A27 interchange in Breda, stores it in and serves a dashboard with a predicted NO2 level and exceedance risk from a trained regression model. Built as the final project of a System Design and Cloud Platforms elective course.
During this 2 weeks company project the objective was to uncover patterns between terminal deployment times, locations, and occurrences of terminal fraud with the absence of direct fraud data. The deliverable was a user-friendly Power BI report designed to support further analysis.
During my school years I had the oportunity to work on both the Unitree go2-W and on the Unitree G1 humanoid robot. More on these projects can be found on my github.
A lightweight, self-hostable personal recipe book for saving recipes generated with LLMs. Recipes are pasted in as Markdown and committed straight to GitHub, with search by name or ingredient, a random pick, and a shopping list that combines ingredients across recipes.
02 / Toolkit
03 / WRITINGS
04 / About
I'm a 3rd year data science student at Breda University of Applied Sciences.
I have a passion for computer vision, Natural Language Processing and data analyzis. I'm fascinated by the challenge of building systems that can sense, think, and act in the real world.
Currently deepening my knowledge in MLops, I enjoy working on hands-on projects that push me to think critically and solve complex problems. When I'm not in the lab or behind a screen, I really enjoy playing chess.
Production MLOps, statistics, and with data store tools and concepts.
A data science / ML engineering internship or data analyst or translator roles where I can ship and measure real systems.
06 / Contact
Open to internships and research collaborations. The fastest way to reach me is by email.