While working at Abbott, I was given a chance to explore all the data generated by the assembly lines. There I implemented a machine learning model to classify the chip protrussion since the off-the-shelf software was already generating labels for the images.
I was able to achieve better accuracy using only 3 out of the 4 images the off-the-shelf software used at one tenth the resolution. Also, in the process I discovered more data needed to be collected at this point in the assembly line to fully compare the current solution with my model's performance.
After showcasing the value ML could bring to the team I was tasked with developing a model that could identify several features from assembly line images and pair them with data stored in the R&D department.
This model was deployed to production as a docker container. Finally, I worked on a third model to approximate the location of a chemical pouch. Due to lockdown this model did not reach production.
Working from home, I developed a graphical user interfase for an in-house statistical analysis tool. This project gave me the opportunity to explore all sorts of data manipulation techniques, and plotting libraries.
The experience at Abbott was key to realize what I want for my future, and I am very grateful for the freedom and trust I was given by the team.
“Manuel is very self-driven to learn new technical skills and apply his learnings to improve project direction and outcome. His software development support has been invaluable for improving our data science capabilities.”