Help & Supportcareers@gravityer.com
Data Scientist | Machine Learning Engineer & Researcher
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Machine Learning researcher and engineer who specializes in the design, evaluation, and productization of deep learning models. BS in Electrical Engineering, MSc. in Data Science, Fulbright Alumni.
New York University
Master of Science in Data Science, Data Science
January 1, 2015 – January 1, 2017
Pontificia Universidad Católica Argentina 'Santa María de los Buenos Aires'
Electric Engineer, Digital Comunications / Control Systems
January 1, 2007 – January 1, 2013
Just Mobile Security
Head of AI
January 1, 2026 – Present
Horus Prosthetics
Principal Data Scientist
January 1, 2024 – December 1, 2025
Hybrid
Goby Lab
ML Engineer & Partner
February 1, 2022 – December 1, 2025
Evernia Bio
Technical Co-Founder
December 1, 2019 – February 1, 2022
Sophos
Principal Data Scientist
August 1, 2017 – December 1, 2019
Nueva York y alrededores
Paperspace
Machine Learning Engineer
February 1, 2017 – August 1, 2017
Greater New York City Area
7 Chord
Data Science Intern
June 1, 2016 – September 1, 2016
New York
Emerging Markets Communications
Network Engineer
February 1, 2014 – June 1, 2015
Emerging Markets Communications
GOC Level 2 Operator
October 1, 2013 – January 1, 2014
Emerging Markets Communications
GOC Operator
September 1, 2012 – October 1, 2013
Pontificia Universidad Católica Argentina (UCA)
Teaching Assistant
March 1, 2012 – July 1, 2013
ExxonMobil
Backup Analyst Trainee
February 1, 2012 – June 1, 2012
Buenos Aires, Argentina
Pontificia Universidad Católica Argentina (UCA) / CONICET
Internship at the interdisciplinary program to study atmospheric processes in global change (PEPACG)
April 1, 2011 – December 1, 2011
Buenos Aires, Argentina
SightWalk
December 1, 2016 – Present
When providing walking directions to a destination, web mapping services usually suggest the shortest route (in terms of distance and/or time). The goal of this work is to automatically suggest alternative more enjoyable routes, which might take marginally more time but go through spots that would be more interesting to the user. In order to do that, we define a scoring function that weights each path based on social media data. Afterwards, we propose two new Graph based optimization algorithms and provide a demo UI.
Mutual Information in Variational Autoencoders
November 1, 2016 – Present
Motivated by the development of InfoGAN, we analyze the role of mutual information in variational autoencoders. We experimentally study the behavior of this model when mutual information between the latent code and the generated data is ex- plicitly enforced as part of its loss function. Furthermore, we make an attempt to formalize the role of MI in the VAE objective. We give an interpretation of a lower bound to the MI as the reconstruction error of a dual VAE.
Identifying Reproducibility in Scientific Publications
May 1, 2016 – Present
In this project we analyze if it is possible to determine wether a scientific publication contains within its text some information about its replicability. Our hypothesis is that there is sufficient information embedded in the body of a published paper to recognize if the reported results are reproducible or not. Our question is, do scientist write in a different way when they are presenting flawed or false results? The way the results are presented, the language used and the level of obfuscation when presenting results might lead us to determine the reproducibility of the research.
Probability Estimation for Online Education System
May 1, 2016 – Present
Online education has become one of the main education formats globally. It has many advantages over traditional education like lower costs, scalability and convenience. However it lacks the personalization of traditional education. It is very challenging to design a system that adapts to the different student skills and weaknesses. By training a machine to understand how well a particular student has learned a particular topic, we could choose which is the most appropriate question to show next in order to optimize her rate of learning. Thus we strive to predict as accurately as possible the probability that a student answers correctly any given question given her past performance. If the probability of answering correctly a certain question is very high, the problem would be too easy - showing that problem to the student would be a waste of time. On the other hand, if the problem had a very low probability of success, it would be too difficult, generating frustration without assisting learning. In this project, we focused on generating calibrated probability estimates for the mentioned task and proposing a personalization system based on these probabilities. The goal was to create a strong probabilistic model using novel feature creation and leverage this model to create an effective recommendation system.
Project Funding Success Probability Estimation
September 1, 2015 – December 1, 2015
• Performed feature extraction and engineering from 100,000+ Kickstarter projects dataset and additional data collection through web scraping. • Conducted exploratory data analysis for features. • Built, trained and optimized a predictive model to estimate project funding success probability. Model and hyper-parameter selection was done using a grid search with cross-validation. Logistic Regression, Decision Tree and Random Forests were considered. Reached an AUC of 0.79. • Develop an optimization algorithm to suggest project proposal changes in order to maximize funding likelihood. • Evaluated the model using Log-loss and the proposed recommender system using expected gain in revenue.
Keyboard Acousting Emanations: New Approximation
September 1, 2013 – Present
Electronic devices might emit different unwanted signals from their common use. By the analysis of this signals it might be possible to recover the original information being at a certain distance. In this work we examine the problem of computer keyboard acoustic emanations. The attack implemented will take as only input a 10 minutes sound recording from a user typing English text in a common computer keyboard and will recover almost 89% of the original characters typed. For the development of the attack, different mathematical tools where applied, a combination of statistical analysis, standard machine learning and speech recognition techniques, including wavelets, mel cepstrum features, Hidden Markov Models, linear classification and feedback-based incremental learning. Key Words: Computer Security, Eavesdropping, Acoustic Emanations, Signal Processing, HMM, MFCC, Wavelets, Machine Learning, Privacy.
Rotary Inverted Pendulum Control System
July 1, 2012 – Present
Final Project for "Robotics Lab" course. The objective is to build and control a real system using the learnt theory in this course as well as in Microcontrollers, Theory and Lab of Control Systems and Robotics. The inverted pendulum system is one of the classic and most important problems on Control Systems Engineering. It is an example of a unstable and non-linear control problem. Project goal was to design and build a prototype of an inverted pendulum and then implement a standalone stabilization routine in a microcomputer using discrete signal processing techniques to allow the pendulum to balance itself in a wide variety of scenarios. The core of the project was focused in the analysis and mathematical modeling of the system and its dynamics, as well as in simulations of the system's response to different types of disturbances.
Cultural Fit Analysis
The candidate demonstrates a strong cultural fit for an innovative and challenging environment, particularly in roles requiring deep technical expertise in Computer Vision and Machine Learning. The breadth of personal projects, from 'SightWalk' to 'Keyboard Acoustic Emanations,' showcases intellectual curiosity and a drive to explore diverse technical problems. Experience in startups (Evernia Bio, Goby Lab) and leading AI initiatives (Just Mobile Security, Horus Prosthetics) suggests comfort with fast-paced, high-impact environments. The academic background and teaching assistant role also indicate a collaborative and knowledge-sharing mindset. The target role of 'Computer Vision' aligns perfectly with recent professional experience.
Soft Skills & Operational Fit
The candidate's experience as a Technical Co-Founder and Head of AI suggests strong leadership, problem-solving, and strategic thinking abilities. The descriptions of projects and roles indicate a proactive approach to developing innovative solutions and managing complex technical initiatives. The diverse project portfolio also points to adaptability and a broad interest in applying AI/ML to various domains.