Tiffany Souterre

Senior Developer Advocate Specialist in AI/ML at AWS

Ph.D. in Genetic Engineering

Image profile

Hi,

I lovelearningandsharingwhat I learn. Being able to explain complex concepts into simple words and make it accessible to everyone is a great feeling. I also love explaining things throughdrawingsand comics.Machine learning and data vizualization are my companions.

Conferences & Co

DRAWINGS

GCP Project

Discover the organization of a GCP project at its creation.

See comic

Preemptible instances

Discover what a preemptible instance is.

See comic

Service account

Discover what a service account is useful for.

See comic

Google Cloud Composer

Discover the GCP service Google Cloud Composer and discover what it is used for

See comic

Feat of arms

Dog breed classifier

  • CNN
  • Face detection
  • VGG16
  • VGG19
  • InceptionV3
  • Resnet-50
  • Python

I built an algorithm capable of identifying canine breed given an image of a dog. If the given image features a human, the algorithm identifies a resembling dog breed. In this project, I leverage the power of Convolutional Neural Networks (CNN) for the dog breed classifier.

See project

Train a Smartcab to drive

  • Reinforcement Learning
  • Q-Learning
  • Optimization
  • Modeling
  • Model Tuning
  • Python
  • Algebra
  • Statistics
  • Calculus

I applied reinforcement learning to build a simulated vehicle navigation agent. This project involved modeling a complex control problem in terms of limited available inputs, and designing a scheme to automatically learn an optimal driving strategy based on rewards and penalties.

Source code   •   See project

Customer segments

  • Clustering
  • sklearn
  • PCA/ICA
  • Visualizing Data
  • Feature Selection

I reviewed unstructured data to understand the patterns and natural categories that the data fits into. Multiple algorithms were compared both empirically and theoretically. I made predictions about the natural categories of multiple types in a dataset, then checked these predictions against the result of unsupervised analysis.

See project

Finding donors to Charity

  • Naive Bayes
  • sklearn
  • Regression vs Classification type problems
  • Model Fitting and Prediction
  • Decision Trees
  • Regression
  • Neural Networks
  • Support Vector Machines
  • K-Nearest Neighbors
  • Adaboosting

Factors that affect the likelihood of charity donations were investigated based on real census data. I developed a naive classifier, trained and tested several supervised machine learning models on preprocessed census data to predict the likelihood of donations. The best model was selected based on accuracy, a modified F-scoring metric, and algorithm efficiency.

See project

Predicting housing prices

  • Bias/underfitting & Variance/overfitting
  • Learning Curves
  • Model Complexity
  • Model Tuning
  • Metric Performance
  • Statistical Analysis
  • Cross Validation

I built a model to predict the value of a given house in the Boston real estate market using various statistical analysis tools and identified the best price that a client can sell their house utilizing machine learning.

See project

Titanic survival exploration

  • Data Analysis
  • Data Preprocessing

In 1912, the ship RMS Titanic struck an iceberg on its maiden voyage and sank, resulting in the deaths of most of its passengers and crew. I explored a subset of the RMS Titanic passenger manifest to determine which features best predict whether someone survived or did not survive. To complete this project, I implemented several conditional predictions.

See project

Tools

Classification

Regression

Clustering

  • K-Means Cluster

    • Dimensionality Reduction

    Model Selection

  • Regularization in Regression

    Preprocessing

  • Text Learning Preprocessing