Ravi Parekh, Ph.D

Education

• 

  Udacity

  2021 - 2021

  Nanodegree Data Scientist

  CRISP-DM process Software Engineering coding best practices Data Engineering - ETL, NLP and ML pipelines. Deployment of an ML model in a flask appExperimental Design and Recommendations - concepts, A/B testing, Recommendation Engines and Matrix Factorization Building Recommendation EnginesData Science project - generating a business questions, use CRISP-DM process to answer questions and report outcomes. Python and Github

• 

  Udacity

  2021 - 2021

  Nanodegree Deep Learning

  Neural Network development using Pytorch. Sentiment analysis and pattern prediction. Convolutional Neural Networks including transfer learning (VGG16/19), weight initialisation, auto-encoders and style transfer. Project in landmark classification from image analysis.Recurrent Neural Networks and LSTM implementation, hyperparameter selection, embedding (word2vec) and sentiment prediction. Generative Adversarial Networks including Deep Convolutional (DC) GANs. Pix2Pix and CycleGan… Show more Neural Network development using Pytorch. Sentiment analysis and pattern prediction. Convolutional Neural Networks including transfer learning (VGG16/19), weight initialisation, auto-encoders and style transfer. Project in landmark classification from image analysis.Recurrent Neural Networks and LSTM implementation, hyperparameter selection, embedding (word2vec) and sentiment prediction. Generative Adversarial Networks including Deep Convolutional (DC) GANs. Pix2Pix and CycleGan with Pytorch. Project on new face generation from existing faces using a DC-GAN. Model deployment in AWS Sagemaker. Hyperparameter tuning, and endpoint generation. Deploying a web-app to interact with AWS endpoint. PythonAWS SagemakerPytorch Show less

• 

  Udacity

  2021 - 2021

  Nanodegree Natural Language Processing

  Building NLP Pipelines - Text extraction from disparate sources (HTML, XML, others).Naive Bayes for classification. Part of speech tagging with Hidden Markov models. Lookup tables, Stemming and lemmatization. bigrams, n-grams and viterbi. Feature extraction for bag of words, TF-IDF, One-hot encoding, word embeddings and embeddings for deep learning.Topic modeling with Latent Dirichlet Analysis. Sentiment analysis using recurrent neural networks (RNNs). Deep learning and… Show more Building NLP Pipelines - Text extraction from disparate sources (HTML, XML, others).Naive Bayes for classification. Part of speech tagging with Hidden Markov models. Lookup tables, Stemming and lemmatization. bigrams, n-grams and viterbi. Feature extraction for bag of words, TF-IDF, One-hot encoding, word embeddings and embeddings for deep learning.Topic modeling with Latent Dirichlet Analysis. Sentiment analysis using recurrent neural networks (RNNs). Deep learning and attention. Machine translation project using RNNs. Building RNNs from input to output using multi-layers, bidirectional layers, time-distributed and dense layers, batch normalization. Voice user interfaces and speech recognition. Audio signal analysis and feature extraction (spectrograms, Mel-frequency cepstral coefficients). Deep neural network speech recognizer on LibriSpeech audio dataset - Creating multi-layer neural networks with CNNs and RNNs to convert audio waveforms to text. Show less

• 

  Udacity

  2021 - 2021

  Nanodegree Machine Learning Engineer

  Software Engineering and Object Oriented Programming in PythonVersion Control (Git)AWS Sagemaker - Model Training, Hyperparamter Tuning and DeploymentAWS Project Deployment - Sentiment Analysis (NLP)Population Segmentation, Fraud Detection, Time-series forecastingReviewed Project in Sagemaker - Plagiarism Detection: Feature engineering, training and deployment using S3 and endpoints. Capstone project on Starbucks dataset - Customer classification based on offer… Show more Software Engineering and Object Oriented Programming in PythonVersion Control (Git)AWS Sagemaker - Model Training, Hyperparamter Tuning and DeploymentAWS Project Deployment - Sentiment Analysis (NLP)Population Segmentation, Fraud Detection, Time-series forecastingReviewed Project in Sagemaker - Plagiarism Detection: Feature engineering, training and deployment using S3 and endpoints. Capstone project on Starbucks dataset - Customer classification based on offer redemption.AWS Sagemaker, Python, (Pandas and Sci-kit Learn) Show less

• 

  Loughborough University

  2009 - 2014

  Master of Engineering - MEng Chemical Engineering

  Activities and Societies: Loughborough Students Union (LSU) Union Councillor 2010/11, LSU Chemical Engineering Treasurer 2010/11

• 

  Loughborough University

  2014 - 2018

  Doctor of Philosophy - PhD Chemical Engineering

  Nonlinear State Space Model Predictive Control (Pharmaceutical Crystallization, Batch and Continuous)Global Linearization with State-FeedbackMechanistic ModellingMulti-Objective Optimisation (Stochastic, NSGA-II)Matlab, Python

• 

  Udacity

  2021 - 2021

  Nanodegree Predictive Analytics for Business

  CRISP-DM analytical framework.Data Wrangling - cleaning, formating, blending.Linear regression, ETS and ARIMA modelling.Binary classification - logistic regression and decision trees.Non-binary classification - decision trees, random forests, boosted models.A/B testing - randomised and matched pair designs.Time series forecasting.Segmentation and clustering - PCA, K-centroids clustering, validation.Peer-reviewed projects for all topics listed above.Alteryx… Show more CRISP-DM analytical framework.Data Wrangling - cleaning, formating, blending.Linear regression, ETS and ARIMA modelling.Binary classification - logistic regression and decision trees.Non-binary classification - decision trees, random forests, boosted models.A/B testing - randomised and matched pair designs.Time series forecasting.Segmentation and clustering - PCA, K-centroids clustering, validation.Peer-reviewed projects for all topics listed above.Alteryx and Tableau. Show less