Waqar Arshad

Start-up working on visual anomaly detection using deep learning. MLops/Backend engineering to help launch the version 1 of the product, setting up testing strategy, continuous machine learning, data version control, containerization, continuous integration and deployment, shell scripting, over the air updates. Managed and mentored two employees. Tools: Python, PyTorch, OpenCV, Docker, AWS, OpenVINO, GitHub, CML/DVS,

Forecasting cloud movement from satellite cloud images using deep learning. Main responsibilities involved data wrangling and pipelining, image processing using Skimage, optical flow methods using OpenCV, researching state of art deep learning networks for ideas and comparison, designing and optimizing deep networks (ConvLSTMs, AutoEncoders). Tools: Python, Tensorflow, Keras, Numpy, Opencv, Skimage, Matplotlib, Azure, Jira

Approximate Computing for Speech Recognition in deep learning applications. Designed a novel Modified-MFCC based feature extractor that increase speech recognition accuracy (keyword spotting) in noisy conditions. Implemented state-of-art deep networks for keyword spotting (TCNs, DS-CNNs) and reduced memory footprint through self supervised quantization and computation requirements through approximate computing techniques. Tools: Python, Tensorflow, Numpy, Linux, CUDA, Bitbucket(Git), Bash, Command Line

Research into AI start ups for the newly launched technology venture. Designed and maintained databases. Tools: MS-Access, VBA, SQL

Implemented machine learning (XGBoost, SVR, ARIMA) and deep learning (LSTMs, CNNs) models to predict future power generation from solar and wind energy farms using weather time series. Data wrangling and data visualization. Tools: Python, Pytorch, Pandas, Numpy, Scikit-learn, Gitlab, Linux, Slurm

Consultant to multiple clients helping them develop big data solutions and deploying them using Amazon Web Services. Tasks involved data storage, pre-processing, feature extraction, ETL, data modelling, visualization and using machine learning methods (Regression, Classification, Decision Trees, PCA, KNN) to get insight and drive business impacts. Tools: Python, SQL, Databases, Hadoop, Spark, AWS, Tableau, Bash, MLib

Used machine learning (PCA, RF, CART) and neural networks (MLPs) to predict drinking water quality from soil data. Data collection and analysis, research into state of art methods. Tools: Matlab, C++, Excel

Machine Learning, Deep Learning, Signal Processing, Communication Systems

Communication Systems, Probability/Stochastic Systems, Linear Algebra, Calculus, Differential Equations, Signal Processing, Electrical Circuits, Computer Networks.