Publication
PyTorch vs. TensorFlow: Differences, Performance, and How to Choose
https://www.toptal.com/developers/deep-learning/pytorch-vs-tensorflow
Nicolas is available for hire
Hire NicolasNicolas Piro
Verified Expert in Engineering
Data Scientist and AI Developer
Zürich, Switzerland
Toptal member since January 21, 2022
Bio
Nicolas is a groundbreaking data scientist with 18 years of experience in data science, software, and R&D in big corporations (Sony, Expedia), healthcare (hospitals), startups, and academia. His areas of expertise include ML, software engineering, signal processing, computer vision, NLP, and big data technologies. Nicolas is a genuinely curious, rigorous, and analytical person with excellent communication skills and a knack for explaining complex concepts to diverse audiences.
Portfolio
SONY
Deep Learning, Computer Vision, CMOS Image Sensors...
Expedia
Python, Spark, Machine Learning, Apache Hive, SQL, Teradata, Presto, Databricks...
Miraex
Python, C, TensorFlow, Data Science, Decision Trees, Dimensionality Reduction...
Experience
- Physics - 17 years
- Jupyter Notebook - 12 years
- Machine Learning - 11 years
- Python - 11 years
- Scikit-learn - 10 years
- SQL - 6 years
- TensorFlow - 5 years
- Spark - 4 years
Preferred Environment
MacOS, Jupyter Notebook, Linux, Amazon Web Services (AWS), Spark, Scikit-learn, Python, PyTorch, C++, Scala
The most amazing...
...thing I've developed is a novel method to help cardiologists visualize the blood flow in the heart and better diagnose certain diseases.
Work Experience
Senior Machine Learning Research Engineer
2021 - 2026
SONY
- Developed a low-latency hand-tracking system using deep learning and computer vision algorithms, deployed in real time in an AR headset system. Used conventional image sensors, event cameras, and depth sensors.
- Built a real-time high-speed 3D sensing system for mobile face recognition. Used event-based sensors for low latency and fast performance.
- Developed a sensor simulation framework and a deep-learning-based processing pipeline to demonstrate a new dual parallel exposure image sensor for HDR and low light imaging, proving better performance than conventional sensors.
- Built a video frame prediction neural network architecture to predict future video frames to compensate for latency in augmented reality head-mounted displays.
- Invented SLAM processing pipeline for hyperspectral sensors on mobile phones.
- Authored 4 patents, mentored junior engineers, delivered customer demos, led projects and teams, managed small engineer teams, and performed reporting duties.
Technologies: Deep Learning, Computer Vision, CMOS Image Sensors, Artificial Intelligence (AI), Transformers, Machine Learning, PyTorch, Open Neural Network Exchange (ONNX), Deployment, C++, Python, Computational Photography, Robotics, Hand Tracking, Augmented Reality (AR), HDR Photography, Data Engineering, Embeddings from Language Models (ELMo), Fully Sharded Data Parallelism (FSDP), Large Language Models (LLMs), Multimodal Models, Sequence Models, Time Series Forecasting, Hugging Face, Supervised Fine-tuning (SFT), Tokenization, Sensor Data, Telemetry
Senior and Lead Data Scientist
2016 - 2021
Expedia
- Built an application to analyze competitor product pricing data and optimize Expedia prices in real time.
- Developed an algorithm to predict user intent in the Expedia Partner Solutions platform and optimize the ranking of the items in the website, which optimized the CTR by over 10%.
- Built experimentation frameworks and analytics tools to monitor and report on the performance of various data science projects in production.
- Developed a prototype machine translation system to translate user reviews in Expedia's main site.
Technologies: Python, Spark, Machine Learning, Apache Hive, SQL, Teradata, Presto, Databricks, Amazon Web Services (AWS), Scikit-learn, TensorFlow, Qubole, Data Science, Decision Trees, Dimensionality Reduction, Gradient Boosting, K-means Clustering, Linear Regression, Logistic Regression, Pandas, Random Forests, Support Vector Machines (SVM), Time Series, Time Series Analysis, Data Scientist, Spanish, AI Agents, Agentic AI, Statistical Learning, Data Engineering, Embeddings from Language Models (ELMo), Fully Sharded Data Parallelism (FSDP), Sequence Models, Time Series Forecasting, Tokenization, Sensor Data, Telemetry
CTO
2015 - 2017
Miraex
- Created data analytics tools and machine learning pipelines to detect anomalies in vibration sensor signals.
- Developed hardware control software for optical and electronic equipment.
- Performed business development activities: pitched in startup competitions, approached and interacted with investors and customers, negotiated pilot projects, and promoted the business in technology fairs and startup events.
Technologies: Python, C, TensorFlow, Data Science, Decision Trees, Dimensionality Reduction, Gradient Boosting, K-means Clustering, Linear Regression, Logistic Regression, Pandas, Random Forests, Support Vector Machines (SVM), Time Series, Time Series Analysis, Data Scientist, Spanish, Statistical Learning, Data Engineering, Internet of Things (IoT), Time Series Forecasting, Sensor Data
Research Scientist
2011 - 2016
EPFL
- Performed cutting-edge research to develop the most sensitive vibration sensors in the world, using optical and nanomechanical technics combined with advanced signal processing.
- Led a team of junior researchers and students, supervising and mentoring them during their projects.
- Performed university teaching and outreach activities.
Technologies: Python, Signal Processing, Machine Learning, Optics, Automation, University Teaching, Decision Trees, Dimensionality Reduction, Gradient Boosting, K-means Clustering, Linear Regression, Pandas, Random Forests, Support Vector Machines (SVM), Time Series, Time Series Analysis, Data Scientist, Spanish, Statistical Learning, Time Series Forecasting, Sensor Data
Experience
Machine Learning App to Predict Failure in Industrial Machines
http://www.miraex.comAn application I developed to analyze data from sensors installed in industrial machines, detect anomalies, predict system failure, and alert operators and maintenance teams. This app is a key component of a predictive maintenance product provided by Miraex.
Automatic Review Translation System for Expedia.com
A prototype of a machine translation system I developed to automatically translate user reviews from English to French, German, Italian, and Spanish on the Expedia.com website. The system was approved for production use.
Pricing Engine for an eCommerce Platform
A pricing engine I built for an eCommerce travel platform. The system used machine learning techniques to automate pricing decisions in real time. It included a real-time analytics platform to test the performance of pricing decisions.
Computer Vision Application to Automate the Alignment and Quality Control of Radiotherapy Machines
An application I built for the radiotherapy department of the main hospital in Barcelona to automate the alignment and quality control of radiotherapy machines. The hospital used the system in production to ease the routine controls in the radiotherapy service.
Education
2006 - 2011
Ph.D. in Applied Physics
Polytechnic University of Catalonia (BarcelonaTech) - Barcelona, Spain
1999 - 2004
Master's Degree in Physics
University of Balearic Islands - Palma de Mallorca, Spain
Skills
Libraries/APIs
Scikit-learn, TensorFlow, Pandas, PyTorch
Tools
PyCharm, Qubole, MATLAB, Open Neural Network Exchange (ONNX)
Languages
Python, C, SQL, C++, Scala
Frameworks
Spark, Presto
Platforms
Jupyter Notebook, MacOS, Databricks, Linux, Amazon Web Services (AWS)
Paradigms
Automation, Anomaly Detection
Storage
Apache Hive, Teradata
Other
Research, Data Processing, Machine Learning, Physics, Statistics, Mathematics, Data Analysis, Optics, Data Science, Decision Trees, Dimensionality Reduction, Gradient Boosting, K-means Clustering, Linear Regression, Logistic Regression, Random Forests, Support Vector Machines (SVM), Time Series, Time Series Analysis, Data Scientist, Spanish, Statistical Learning, Deep Learning, CMOS Image Sensors, Artificial Intelligence (AI), Multimodal Models, Sequence Models, Time Series Forecasting, Sensor Data, Software Development, Programming, Signal Processing, University Teaching, Computer Vision, AI Agents, Agentic AI, Transformers, Computational Photography, Hand Tracking, Augmented Reality (AR), HDR Photography, Data Engineering, Fully Sharded Data Parallelism (FSDP), Internet of Things (IoT), Large Language Models (LLMs), Hugging Face, Supervised Fine-tuning (SFT), Tokenization, Telemetry, Qt 4, Natural Language Processing (NLP), Translation, Generative Pre-trained Transformers (GPT), Deployment, Robotics, Embeddings from Language Models (ELMo)
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring