Jakob is an Engineer turned Data Scientist and Machine Learning Engineer, with proficiency in Python and its frameworks. He has the ability to transform a sea of data into actionable insights that can have an impact on businesses and projects- from building predictive models for various types of cancer detection, to setting-up consumer facing models in the transportation and entertainment sectors, as well as the financial industry. In his role as an Educator in the Data Science and Data Analytics programs, he has shown great communication and mentorship skills, having encouragingly guided his students to achieve high-level technical skills and great professional performance in the industry.
Acted as a part of the Data & Insights team at this 300M+ funded hyper-growth scale-up, building out the AI system on the platform. Created a satisfaction model from scratch to identify unsatisfied users on the platform. The model was deployed to AWS and both Looker and Streamlit dashboards were used to communicate the output to various stakeholders.
Developed a Grafana dashboard with RedShift data source was used to visualize the AB tests on the productionalized model.
Created a model that significantly improved the company's capacity to enhance user experience while saving costs. Using Hugging Face, I built out an NLP model in record time to identify the Language, Sentiment and Topic of user feedback despite limited amount of data.
Built, using Django and TensorFlow, a machine-learning grading system which automatically evaluates open-ended questions in schools.
Acquired data, through qualitative and quantitative analysis, as well as unbiased user interviews, to guide the engineering process.
Built scalable full stack Deep Learning system applied to qualitative data (text); lead the experimentation and development of scalable NLP models to deliver production-ready solutions.
Developed machine learning models for cancer detection, based on a novel cancer detection approach to improve model performance; the models showed improved sensitivity by 34%, 21% and 15% for breast, colorectal and pancreas cancers respectively.
Collaborated with engineers, designers and developers to build Web platforms for a range of a clients.
Mentored junior engineers as an Educator in the Data Science and Data Analytics programs; guided his trainees through the industry's best practices by translating concepts and terminology from data science, machine learning and data analytics into easy-to-digest curriculum.
Strategically planned and delivered assigned projects and the curriculum learning progressions.
Effectively trained students which have managed to take on Director positions at start-ups, join top 3 world ranked Universities and work for the financial institutions of Wall Street.
Collected and analyzed 2.5 million social media posts to identify key public opinion trends and patterns, while providing valuable analytical insight to help stakeholder’s decisions.
Conducted behavioral segmentation with quantitative and qualitative data using clustering techniques, by applying Unsupervised Machine Learning techniques; automated the data mining processes with Python for faster and more accurate results.
The present findings were published in several national Spanish newspapers.
Jakob has worked on improving a machine learning model for early cancer detection. He applied state-of-the-art machine learning algorithms to a publicly available dataset and attempted to improve the results obtained by a research team at Johns Hopkins University. His ultimate goal was to sell the machine learning implementation to Thrive Earlier Detection.
Thrive was a healthcare start-up dedicated to incorporating earlier cancer detection into routine medical care. Thrive went on to receive FDA approval and has been acquired by Exact Sciences for $2 billion.
Jakob built a predictive model for taxi fares using a large dataset. He applied machine learning techniques to a publicly available cab rides dataset for New York City, which includes over 1.1 billion rides between 2009 and 2015. The main objective was to display the implementation's cost-effectiveness and showcase the work to the community.
For his Master's thesis, Jakob built an end-to-end deep learning system to predict whether distant stars had habitable planets orbiting them. His objective was to evaluate the capabilities of Deep Learning algorithms, more specifically Convolutional Neural Networks (CNNs), with a regression approach, on predicting physical stellar parameters from spectrograms.
Education
Master degree in Science, Industrial Engineering and Data Science
