About

Hi! Felipe here. Machine learning researcher and engineer, specializing in design and evaluation of deep learning models. Currently working independently under Goby Lab, where we strategize and build data solutions for mid and small sized companies.

Before that I worked at the intersection of computer vision and microscopy building algorithms for estimating biomass of microorganisms, designed and built deep learning models for applications in cybersecurity @ Sophos, and trained state of the art models for stress-testing Paperspace platform capabilities.

Coming from the world of Electrical Engineering, satellite communications and building robots (Universidad Catolica Argentina), I received my Master’s degree in Data Science at the Center for Data Science at New York University. While there I focused my research on adversarial learning applied to generative models.

Curious by nature, I enjoy gardening, yoga, rock-climbing, books, and spending time outdoors. Other interests include, nature inspired solutions, regenerative agriculture, metagenomics, bio-based materials, myths, and eastern philosophy.

Some projects I’ve worked on’:

• Green-ML resources. [github-repo]

• Felipe N Ducau, Ethan M Rudd, Tad M Heppner, Alex Long, Konstantin Berlin. SMART: Semantic Malware Attribute Relevance Tagging. 2019. [arxiv-preprint] [CAMLIS Presentation (video)]

• Ethan M Rudd, Felipe N Ducau, Cody Wild, Konstantin Berlin, Richard Harang. ALOHA: Auxiliary Loss Optimization for Hypothesis Augmentation. 28th USENIX Security Symposium 2019. [paper]

• Richard Harang and Felipe Ducau. 2018. Measuring the Speed of the Red Queen’s Race; Adaption and Evasion in Malware. Presented at BlackHat USA 2018. [pdf] [talk] [slides]

• Felipe Ducau, under the advice of Stanislas Lauly and Kyunghyun Cho. 2017. SEQ2SEQ Professor Forcing [pdf]

• Felipe Ducau and Maria Elena Villalobos Ponte. 2016. CWord: Incorporating Larger Context in Neural Conversation Model. [pdf] [source]

• Felipe Ducau, Maria Elena Villalobos Ponte, Sebastian Brarda. 2016. SightWalk: Automatic Generation of walking paths from social media. [pdf] [demo] [source]

  Recipient of the 2016 Capstone Project of the Year award - NYU Center for Data Science.

• Felipe Ducau and Sony Trenous, under the advice by Joan Bruna. 2016. infoVAE: Mutual Information in Variational Autoencoders. [pdf] [source]

Blog Entries

• Adversarial Autoencoders (with Pytorch)
• Recurrent Neural Networks - Building the intiution