Currently employed in the Siri Information Intelligence, AIML organisation, part of a dynamic team dedicated to leveraging ML in an innovative platform, collaborating closely with cross-functional teams, applying expertise in LLM-based workflows, and rapid iteration with fine-tuning toolboxes to deliver high-quality models meeting stringent performance standards. Translating high-level product goals into actionable data, model, and metrics requirements while maintaining keen awareness of model complexity, power consumption, and overall performance. Proficient in generating value from models by seamlessly integrating them into complex systems, ensuring a cohesive user experience.Activities: Co-author 2 papers & reviewer (Apple internal conferences: WCST, MLS)

Engaged in developing an intelligent robotic perception system focused on detecting and localizing plants for selective spraying, with each target plant being RTK tagged in real-time through diverse sensor inputs. Key responsibilities included crafting efficient algorithms for plant segmentation in the image plane, 3D space localization, and real-time tracking and storage within tree databases. Research and implementation efforts centered on pioneering Deep Learning (DL) models for real-time optical flow estimation within an embedded module, emphasizing linear scaling properties and the creation of a parallelizable architecture capable of on-demand processing while considering context, and tensor similarity using methodologies like Deep matching, CNN, and ConvGRU. Additionally, responsibilities encompassed refining and optimizing state-of-the-art DL models for real-time flare segmentation, as well as deploying and optimizing models on CPU/GPU platforms. The role entailed comprehensive data preparation and engineering, coupled with model development, validation, optimization, and deployment. Further contributions involved the design and implementation of image processing algorithms for soil segmentation, shadow segmentation, and vignetting correction, along with geospatial data analysis.

At DHL, focused on developing and evaluating Machine Learning algorithms to detect rare events within noisy data, employing dimensionality reduction techniques to create meaningful feature embeddings. Responsibilities encompassed the development, training, and testing of Autoencoders (AE), Variational Autoencoders (VAE), LSTM-AE, and other ML algorithms to identify relevant embeddings. Additionally, trained a Doc2Vec model to embed categorical features and compared dimensionality reduction algorithms such as PCA, SVD, and t-SNE, while conducting data exploration and visualization on a graph database. Within iLab Politie, engaged in full-stack development of a proof of concept speech-based virtual assistant named NEXUS, offering detailed briefings to police officers by connecting to the Police database and delivering personalized information. Development tasks included utilizing Vue.js, Flask, and MySQL for the full-stack implementation of NEXUS, alongside conducting a literature review of Deep Learning approaches for natural language queries to SQL queries.For AppsForce, spearheaded the design of a scalable proof of concept in precision agriculture termed Animus, aimed at optimizing crop management by leveraging sensor data such as NDVI from satellite images, IoT sensors, and weather forecasts. Responsibilities entailed building a data pipeline to process remote/multi-spectral sensor data, developing a recommender system for crop management, and demonstrating the value of the proposed approach on a Dutch wheat crop by optimizing nitrogen intake (variable rate, timing), water intake (based on soil moisture and weather forecast), controlling soil acidity, and predicting yield.

During the summers of 2017, 2018, and 2019, I served as a student at ING Belgium, contributing to data analysis, reporting, and task automation initiatives. A significant project involved the development of an information system aimed at monitoring employee performance and objectives, culminating in the presentation of my data visualization Excel tool and findings at the mid-year general meeting held at the Brussels head office. Key responsibilities included crafting a data visualization tool for weekly reporting and business intelligence purposes, as well as designing Key Performance Indicators (KPIs) to evaluate the effectiveness of the new agile organizational system. Additionally, I conducted statistical analysis and built models to support strategic business decisions, while leveraging VBA macros to automate redundant tasks, enhancing efficiency across various operations.