Ypres, Flemish Region, Belgium
Summ.Link is a 1-year-old start-up that uses Machine Learning to innovate the world of HR. I use NLP and ML to accelerate hiring and save time for recruiters. - Researched, implemented and deployed (API) several models (E.g. LLMs) related to resume / job feature extraction and the matching of those. - Developed CI/CD pipelines for model testing and deployment while managing AKS (Azure). - Created a Django backend to interface with the frontend, handling resume upload, triggering the async computation of features and predictions to serve suggested jobs for a given uploaded resume.
Defined AI (formerly DefinedCrowd) is an 8-year-old start-up that combines Crowd-as-a-Service and Machine Learning to accelerate enterprise data training and modelling. I use data science and machine learning to improve annotation quality and speed or reduce cost on the online data labeling platform Neevo. - Supervised the transition to a machine learning based voice activity detector, eliminating the need for Solution Developers to manually set parameters.
Advocate a data-driven culture by including Data Science in early product development and supporting Data Scientists in intra-squad communication. - Researched and implemented (Python library) the automated detection of dishonest contributors without the need for test questions for speech collection and transcription workflows, reducing setup time from 3 weeks to 2 days. - Researched and implemented background noise detection in speech collection workflows (Python library + Nuclio API deployed in Kubernetes), reducing the need for human labeling by 71%.
Predicted human effort & cost needed to collect a dataset (e.g. to improve speech to text), eliminating the risk of underselling a workflow.