Ramtin Rassoli
Verified Expert in Engineering
Deep Learning Developer
Ramtin received his master's degree in applied computing from the University of Toronto in 2018, and since then, he's been working as a data scientist. He has conducted research on extracting behavioral patterns of users and activity recognition in the context of a smart home. At ecobee, his major contributions have been around helping to build their data services and models at a large production scale.
Portfolio
Experience
Availability
Preferred Environment
Jupyter, IntelliJ IDEA, Visual Studio Code (VS Code), PyCharm
The most amazing...
...project I've been involved with is a smart home cloud application with a microservice architecture and several ML components deployed to Google Cloud Platform.
Work Experience
Senior Software Engineer | Tech Lead
BenchSci
- Moved the training infrastructure of our ML team to Google Vertex. This includes experiment tracking, metadata tracking, artifact versioning, hyperparameter tuning, and distributed training.
- Designed and deployed the company's batch inference workflows on Google Dataflow.
- Formed and hired the company's MLOps team with five engineers.
Software Engineer | MLOps
BenchSci
- Developed an end-to-end pipeline for training and deploying production ML models using TensorFlow Extended (TFX).
- Tracked ML experiments and data versioning with MLFlow and DVC.
- Implemented feature engineering and orchestration pipelines with Apache Beam and Google Dataflow.
Senior Data Scientist
Ecobee
- Helped to build Ecobee's data platform team and the ML infrastructure for model training and validation.
- Developed light-weight acoustic event detection models to detect different audio events on IoT devices.
- Oversaw the data engineering processes for Ecobee's home monitoring initiative.
Data Scientist
ecobee
- Developed supervised LSTM models to detect acoustic events in homes.
- Implemented batch and streaming data pipelines using Apache Beam and Google Dataflow.
- Built unsupervised clustering models to extract user schedules and behavioral patterns.
- Designed and implemented supervised models to predict occupancy and activity type at home using Keras and scikit-learn.
- Deployed highly scalable microservices to Google Kubernetes Engine in Golang.
Experience
Home Occupancy Prediction
The data cleaning pipelines were written in Java using Apache Beam and deployed to Google DataFlow. The model was written in Python3 using Sklearn and Pandas and is served as a GCP Cloud Function to predict home occupancy likelihood in the home monitoring and security context.
The end-to-end workflow is also scheduled and monitored by Apache Airflow and deployed to Google Cloud Composer.
Education
Master's Degree in Applied Computing
University of Toronto - Toronto, Canada
Bachelor of Science Degree in Information Technology Engineering
Sharif University of Technology - Tehran, Iran
Certifications
Deep Reinforcement Learning Nanodegree
Udacity
Deep Learning Specialization
Coursera
Skills
Libraries/APIs
Scikit-learn, Keras, TensorFlow, PyTorch
Tools
Apache Beam, Cloud Dataflow, PyCharm, IntelliJ IDEA, Jupyter, Google Kubernetes Engine (GKE), Jira, Confluence, Bazel
Languages
Python, Python 3, Go, Java
Paradigms
Data Science, Microservices, MapReduce
Platforms
Google Cloud Platform (GCP), Kubernetes, Vertex AI, Visual Studio Code (VS Code)
Storage
MySQL, PostgreSQL
Other
Google Pub/Sub, Deep Learning, Google Cloud Functions, Deep Reinforcement Learning, Data Visualization, Machine Learning
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