Dwayne Fernandes

Major responsibilities include mining fashion-items data. This is achieved by building an API for the detection and extraction of fashion items from an image and grouping the same items together. The API is then supplied with images that could be linked to a user's social media account to create a fashion profile for that specific user.

Developed an API for extraction of fashion-items from an image using object-detection neural models for mining of fashion data in Pytorch
Increased performance in grouping and matching similar fashion-items using K-means clustering algorithm
Built an end-to-end pipeline for creating a fashion-profile for a user by linking the fashion-API to the users instagram account using Python

Developed a Reinforcement agent for automatic cultivation of hydroponic crops by controlling growth factors like temperature, light intensity, nutrient compositions, irrigation, CO2, etc. using Python
Built an interactive dashboard using Plotly for real time logging and monitoring of the plant growth cycle and behavior/performance of the agent
Achieved an accuracy of over 91% for plant disease detection of 30 fungal diseases using CNN, OpenCV, and Keras-TensorFlow
Deployed the end-to-end system on AWS using a serverless architecture

Developed an Intelligent Trading Bot for automatic trading of Bitcoin using Time-Series forecasting and Sentiment analysis of bitcoin tweets
Enhanced forecasting results by using neural models in Tensorflow, Python by reducing the mean squared error

Analysed Financial Data feeds using Exploratory Data Analysis in Python
Successfully identified clusters of users to offer loans based on historical data
Built dashboards for interactive visualizations of the analysis using Javascript and NodeJS

Developed an API for extraction of fashion-items from an image using object-detection neural models for mining of fashion data in Pytorch
Increased performance in grouping and matching similar fashion-items using K-means clustering algorithm
Built an end-to-end pipeline for creating a fashion-profile for a user by linking the fashion-API to the user's Instagram account using Python

Developed a Reinforcement agent for automatic cultivation of hydroponic crops by controlling growth factors like temperature, light intensity, nutrient compositions, irrigation, CO2, etc. using Python
Built an interactive dashboard using Plotly for real-time logging and monitoring of the plant growth cycle and behavior/performance of the agent
Achieved an accuracy of over 91% for plant disease detection of 30 fungal diseases using CNN, OpenCV, and Keras-TensorFlow
Deployed the end-to-end system on AWS using a serverless architecture

Developed an Intelligent Trading Bot for automatic trading of Bitcoin using Time-Series forecasting and Sentiment analysis of bitcoin tweets
Enhanced forecasting results by using neural models in TensorFlow, Python by reducing the mean squared error

Analysed Financial Data feeds using Exploratory Data Analysis in Python
Successfully identified clusters of users to offer loans based on historical data
Built dashboards for interactive visualizations of the analysis using Javascript and NodeJS

The job is focused on applying machine learning in the cultivation of hydroponic crops. Major responsibilities include developing a Reinforcement agent for the automation of cultivation of various crops my monitoring and manipulating critical factors of the controlled environment such as light intensity, nutrient compositions, water, CO2, temperature etc. Developed a disease detection module based on CNN achieving an accuracy of over 92% on over 30 different fungal diseases.

Developed an auto trading bot for Bitcoin that trades based on forecast and sentiment of bitcoin tweets. The forecasting model was built on a neural architecture to attain very small MSE from the actual price while the sentiment analysis was doing using NLTK and neural nets.

Analysed Financial Data feeds using Exploratory Data Analysis in Python. Successfully identified clusters of users to offer loans based on historical data.
Built dashboards for interactive visualizations of the analysis using Javascript and NodeJS