Tech Lead

• Partnered with Senior Data Scientist to create an Advanced Fire Detection Model (Patent Pending) for the multiple Raspberry Pi nodes. • Researched and implemented fire detector model architecture to produce 90% accuracy, increased prediction speed from 1000ms to 90ms with +-1% loss and converted model from 4MB to TF lite 1MB and loaded model to Raspberry pi.• Published "Installing TensorFlow 2.3.0 for Raspberry Pi3+/4 (Debian Buster)" on Medium|ITNEXT reached over 7k views.• Designed error analysis plots as a prediction evaluation metric for long sequences of classifications.• Researched and implemented feature importance and model compression techniques (Post-quantization and pruning)

Thermal CCTV Body Temperature Monitoring for COVID Project.Collaborated with a senior data scientist to develop a face detection and temperature detection system which alerts users that have a fever.

• Launched a data-capturing project tracking click events of mobile and web users. Designed, developed and maintained a unified event tracking schema. Utilised Redshift and Presto to perform user segmentation and funnel analysis through click analysis, enhancing user retention rates.• Worked with the Regional Head of Communications in preparation of the “Year In Review 2018” report. Found the significance of a mobile-first strategy through analysing the volume of traffic, average basket size and frequency.

• Worked with the Battalion 2nd Commanding Officer to redesign and automate network switch configurations for a military satellite vehicle to enable fast uplink and communications.• Streamlined VSAT operational processes by abstracting complexities like encryptor configurations from users, reducing the learning curve for the unit’s NSmen (Reservist) and achieving an 81% (40% increase) of overall satisfaction among clientele.• Awarded Best Soldier in HQ Guards Unit by COL Chief Guards Officer