Professional Experiences

I worked on object classification for power poles and relevant assets recognition which can be used to knowledge acquisition from spatial data and to create an annotated map.

My duties including but not limited to research and development on the following tools and high performance computing methods using RGB data:

• data-preparation
• data structures and manipulation
• visualisation
• classifier training
• evaluation

Classification methods including but not limited to:

• haar cascade
• Deep and Convolutional Neural Networks such as OverFeat, Tensorflow, Caffe, Cuda-ConvNet, Darknet and Keras (using Tensforflow as back-end) on CPU-only and GPU-based systems (on AWS EC2 instances)

About this project: While banks strive to make their services widely available through branches and automated teller machines, the cost of distributing cash to these locations can be high. But failure to replenish them in a timely fashion can quickly lead to customer dissatisfaction.
Consensio is a software tool that uses advanced analytic to determine the optimal replenishment cycle for every point in the cash distribution network. This gives the banks, as well as large retailers, the greatest possible flexibility to manage their cash availability in its network at the lowest possible cost by forecasting cash usage patterns and optimising delivery schedules. Consensio is the only product developed specifically to handle the issues relating to cash management, and is also the only one that manages the entire cash life-cycle as it moves into the retail system and back out for distribution. This can result in further cost savings for all parties involved through sharing resources.