Nicolas V.

Juju core developer

Software developer at the backend team, where we are responsible for building the Keen Eye platform, deployment of health AI applications, image processing, user management and more. Some technical challenges encountered and successfully solved include: • Designed and implemented a native library (using cgo) written in go for performant image reading from python. This included a lot of pprof profiling and performance tuning. • Re-designed and implemented the "tiling" service, which takes a whole slide microscopy im- age (several tens of thousand of pixels height/width) and creates a pyramid that allows multi- resolution display on our web-app. • Re-designed and implemented the service that manages geometrical image annotations (several millions per image) that are displayed in the web app using dimensionality-reduction techniques on the database thanks to PostGIS. I also took an active role in the technical teams reorganization, development process and practices (pair programming, merge requests, TDD), methodology and remote culture of the company. Technical stack: - Go, C++ - Image processing, OpenCV - MongoDB, PostgreSQL, PostGIS - DevOps & CI: Kubernetes, Docker, Git, Gitlab, Linux

Software engineer at Groupe d'Imagerie Neurofonctionelle (http://www.gin.cnrs.fr) developing applications for medical (neurological) image processing, storage and managing. In charge of software development for the neuroimaging protocol of i-Share (http://www.i-share.fr/). Technical stack: - Java/ Java EE, Spring framework (Boot, Data) - Matlab - PostgreSQL - XNAT Publications: Tzourio-Mazoyer, Nathalie, Laurent Petit, Laure Zago, Fabrice Crivello, Nicolas Vinuesa, Marc Joliot, Gael Jobard, Emmanuel Mellet, and Bernard Mazoyer. "Between-hand difference in ipsilateral deactivation is associated with hand lateralization: fMRI mapping of 284 volunteers balanced for handedness." Frontiers in human neuroscience 9 (2015).

Working in group GEOSTAT. Subject: “Biologically realistic coding efficiency in the auditory cortex versus wavelet analysis”. Technical stack: - Matlab - Java - C Publications: Khalid Daoudi, Nicolas Vinuesa. An analysis of psychoacoustically-inspired matching pursuit decompositions of speech signals. International Conference on Natural Language, Signal and Speech Processing, Dec 2017, Casablanca, Morocco.

Internship in group PAROLE. Subject: “Dealing with automatic classification uncertainty in training acoustic models for speech recognition”. - Hidden Markov Models (HMM) for speech recognition, Speaker recognition. Technical stack: - Matlab - Perl / Bash - HTK Publications: Jouvet, Denis, and Nicolas Vinuesa. "Classification margin for improved class-based speech recognition performance." In Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on, pp. 4285-4288. IEEE, 2012.

Aardvsys was a high-end audio startup that produced amplifiers and Audio over Ethernet products. As a technician and junior engineer I was involved with the design, test, manufacturing lifecycle of D-class, ethernet-enabled (using the CobraNET protocol) audio amplifiers.