Piero Ferrarese
Verified Expert in Engineering
Data Scientist and Developer
Piero is a data scientist and developer with a computational physics background. Answering crucial business questions by extracting high-level information from raw data is what drives Piero. He enjoys working on complex problems end-to-end, from IoT programming on Rasberry or Arduino down to applications layers using ML and AI models. Piero focuses on predictive analytics and big-data challenges but has vast experience in various data science topics.
Portfolio
Experience
Availability
Preferred Environment
Deep Learning, Machine Learning, R, C++, Predictive Analytics, TensorFlow, Python 3, Arduino, Raspbian
The most amazing...
...application I've built: a Raspberry Pi-based system, transmitting tracking information from fishing vessels, aggregating, and analyzing them in a Kafka Cluster.
Work Experience
CTO
nPhysis srl
- Developed an API that retrieves a comfort index for any skiing slope around the world based on weather conditions and orographic. Forecasting values for the index are supplied with an overview of weather conditions.
- Employed GDAL to manipulate weather models and satellite images, generating tiles and data directly employed within Mapbox GL JS or LeafletJS.
- Led a team of three to design, develop, and launch the comfort-index product. Enforced the usage of scrum and automation from unit tests down to deployment scripts and processes in a CI/CD fashion.
Chief Data Scientist | Co-founder
SciScry GmbH
- Developed ML and DL applications for forecasting and data science applications. Anomaly detection on transactions, recommendation systems for customer support application, forecasting thousands of product/customer combinations through AutoML.
- Led, as the chief data scientist, the design, development, and deployment of apps. Projects included time-series forecasting and analysis, demand planning, capacity planning, anomaly detection, market segmentation, and recommendation systems.
- Developed a time-series forecasting AutoML engine. Customers supply the data and get the best forecasts generated by state-of-the-art ML and DL algorithms.
- Supervised working students who were compiling a master's degree thesis within SciScry. Topics include copula models in time-series analysis, genetic algorithms for model selection, and optimization.
- Ran several data science design sprints to gather requirements and prepare prototypes for exploring innovative solutions to business needs.
Data Science Consultant
Noventi SE
- Developed a reporting dashboard for analyzing anomalies in sales transactions from more than three years of daily records over 2,000 pharmacies.
- Generated insights that take into account POS information and third-party data related to demographics, competitors nearby, and clustering results.
- Generated predictions that can be inspected in detail, thanks to interpretability models, i.e., Shapley values.
- Ran data aggregations by scaling MongoDB on the distributed system for over one billion documents with processing time in the order of minutes.
Data Science Consultant
Automotive Supplier
- Used the time-series platform developed by SciScry to generate forecasts for the different products/customer segments to pre-fill the next business-year demand plan.
- Increased accuracy on average by 20% of the total number of items sold concerning the company business plan.
- Scaled the application on AWS EC2 to run preprocessing, feature engineering, model selection, and forecasts for more than 80,000 product/customer combinations in less than 24 hours.
- Implemented ad-hoc models for handling product and customer combinations with low and intermittent sales volume, increasing the accuracy by 40% for this class of products.
Junior IT Consultant
Adidas AG
- Developed the back end (SAP BW) and front end (SAP BusinessObjects) of a planning application, employed by almost 50 users of the different Adidas business units across the globe.
- Designed a solution to optimize the product portfolio using Monte Carlo simulations to reduce the volatility of demand and prices.
- Automated the process of testing process chains using a Python script to consolidate the same datasets in different environments. This sped up the development time by 30%.
- Executed projects following scrum practices and led a team of five developers during the whole app lifecycle, from the application's design with the product owner to the deployment and knowledge transfer to the support team.
Experience
Wikipedia Page Request Forecast
https://bit.ly/2ZGOSxOThe task is to forecast the requests of a page to trigger an alert when the demand may become too high for servers to be handled. I employed Google Trends data as alternative data to enhance the accuracy of the predictions. The model used is an LSTM with encoder-decoder representation.
The results have been presented at a major German conference for machine learning applications.
My Research Publications
https://bit.ly/2FIcbzUSciScry Time-series Forecasting Engine
https://www.sciscry.aiEducation
Ph.D. in Physics
Georg-August-Universität Göttingen - Göttingen, Germany
Master's Degree in Physics
University of Padova - Padova, Italy
Bachelor's Degree in Physics
University of Padova - Padova, Italy
Certifications
C1 German Language
Sprachschule für Deutsch als Fremdsprache Erlangen
Scrum Master
Scrum Alliance
Skills
Libraries/APIs
TensorFlow, XGBoost, Scikit-learn, Mapbox GL, Leaflet
Tools
StatsModels
Storage
MongoDB
Frameworks
Flask
Languages
Python 3, Python, C++, R, SQL, Excel VBA
Platforms
Amazon EC2, Raspbian, Arduino, Mapbox
Paradigms
Parallel Computing, Distributed Computing, Anomaly Detection, Agile, High-performance Computing (HPC)
Other
Predictive Analytics, Monte Carlo Simulations, Dash, Multiprocessing, Machine Learning, Deep Learning, AWS Cloud Architecture, Satellite Images, Scrum Master, Data Analysis, Statistics, Markov Chain Monte Carlo (MCMC) Algorithms, Multivariate Statistical Modeling, High Energy Physics (HEP), Predictive Maintenance, Google Trends, LSTM Networks, Time Series Analysis, Time Series, Physics Simulations, Weather Research & Forecasting (WRF), Recommendation Systems
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring