Jawahar Babu SELVARAJ

Software Engineer

• Developed software applications in mobile, desktop and web platforms in flutter, which works with motion sensor (MEMS) data.
• Developed an automated testing framework, drastically reducing embedded algorithms’ validation and verification time by almost 85%.
• Designing and maintaining Continuous Integration and Delivery pipelines for software applications.
• Experienced working with International clients worldwide and attending business meetings overseas.

Project Trainee

• Developed a Deep learning image classifier to detect a defective sensor and to automate manual sensor inspection, thereby speeding up the manufacturing process.
• Used the variational autoencoder-based semi-supervised model with image processing techniques to build a state-of-art classifier with an accuracy of 93 %.

Web Development Intern

• Developed a real-time IoT - MQTT dashboard to monitor and control IoT devices, for the application of Home Automation
• Used Angular JS frontend and Express JS API endpoints with mosquitto MQTT broker.

Summer Intern

• I learnt various telecom technologies including but not limited to LTE, GPRS, Fibre optic ommunication with hands on experience.

Carleton University

CGPA: 11/12

Government college of Technology (Anna University)

CGPA: 8.6

Automatic floor cleaning bot

• Developed a minimum usable prototype of an economical vacuum robot using an internal navigation system, which finds its applications in Hospitals, Malls and other public places
• implemented the Incremental D*, a grid search algorithm in python, for path mapping and navigation of the robot, using cost-effective ultrasonic and motion sensors supplanting an expensive lidar. The algorithm was optimised for running on an Arm Cortex A53 microprocessor.

Traffic Sign Classifier in Self-Driving Cars

• Developed a working prototype of a rudimentary self-driving car, which pilots based on a robust image classifier that recognises traffic signs
• Built a ResNet CNN model using TensorFlow, which was trained on a custom dataset with thousands of data on each class of traffic sign.
• Applied image processing techniques like edge detection, thresholding and pruning using openCV to the training and real-time data, enabling the classifier to attain an accuracy of 95%.

Automatic Fire Fighting Robot

• Built a small-scale PoC of a fire-extinguishing robot, which detects fire in its vicinity, navigates towards it and extinguishes it without human intervention.
• Designed a digital circuit, which takes input from IR and ultra-sonic sensors, and actuates the motors for navigation, using the L293D motor driver IC.