Certified Scrum Product Owner (CSPO)

Deep Learning Specialization

C Standard Library

Sensor-Fusion Nanodegree

Build Basic Generative Adversarial Networks 

Advanced Python

Software Testing Foundations: Test Techniques

Learning REST APIs

Advanced C Programming

Learning Python Generators

First Look: Rust

Introduction to Machine Learning in Production

Bachelor of Science (B.Sc.)

Self-driving Car Nanodegree

Master's degree

* AWS S3 and EC2
* Database creation and  administration with MongoDB 
* ETL scheduling and orchestration using Prefect

Senior Data Engineer and Data Scientist 

The ATLAS experiment is located at the European Center for Nuclear Research (CERN) records collision data at a rate of 1 GB/s. Such massive amounts of data need to be carefully processed and filtered to achieve an optimal signal identification. My work focused on the last stages of the signal identification, where it becomes increasingly difficult to disentangle signal from background. Optimizing and studying unsupervised state-of-art clustering techniques it was possible to achieve marked improvement on the signal identification performance.

Doctoral Student

During this brief period of post-doctoral research, I continued conducting data analysis and optimizing algorithms using the petabytes of data collected at CERN. Successfully supervised a bachelor's thesis.

Post-Doctoral Resarcher

Working on implementing a High Beam Assistant (HBA) recognition system with state of the art convolutional neural network techniques that are able run in real time in a vehicle. The HBA optimizes the usage of the high beam during the night giving the driver and other drivers on the road more comfort and safety. The HBA system is being developed and implemented in accordance to ASPICE standards. 

Additionally I am working on a Vision Fail Safe (VFS) system for automotive cameras to detect blockages such as soiling, dust and difficult environment such as low-illumination and glare (due to low-sun or other vehicles at night). Such a system can be used to determine the reliability of the detection algorithms, as well as to improve the decision performance of the control system i.e. an automated lane change system may avoid action if the rear-side cameras are blocked or impaired. Camera self-cleaning systems can also use the VFS algorithm to determine when the camera needs cleaning.

David Sosa

PhD Physics, Machine Learning Engineer, Founder