Afrasiab Ahmad

◦ Deep Learning - Artificial Intelligence: Customized MobileNetV2 pre-trained CNN using transfer learning
with 800 hand gesture images (400 open/400 closed) for binary classification. Achieved 100% accuracy, sensitivity,
precision, recall, and F1 score with hyperparameters: 50 epochs, 0.001 learning rate, 16 batch size.
◦ Applied Computer Vision for AI: Developed real-time hand gesture recognition system using laptop webcam,
image cropping as preprocessing, and binary classification (open hand = START, closed hand = STOP). Deployed
model for industrial conveyor control.
◦ Industrial Variable Frequency Drive (VFD) System: Configured iG5A VFD drive with parameters including
5 sec linear acceleration and 10 sec S-curve deceleration for smooth three-phase motor control. Connected VFD to
relay and Arduino Uno.
◦ Industrial Three Phase Delta Motor: Integrated 0.5 HP, 220V AC, 60 Hz three-phase motor in delta
connection with VFD. Analyzed motor startup, speed control, and emergency stop behavior for industrial
automation.
◦ Firmware Programming in Chip: Deployed trained MobileNetV2 deep learning model onto Arduino Uno
(ATmega328P) using AI edge computing principles. Programmed gesture inference logic directly on embedded chip
without cloud dependency.
◦ Electronic Relay Switching System: Designed opto-coupled electromagnetic relay switching circuit (5 pins:
common, normally open, normally closed). Connected relay between Arduino Uno and VFD for low-voltage logic to
safely control motor power.
◦ Interfacing of Electronic Devices: Interfaced HC-05 Bluetooth module (2.4 GHz ISM band) with Arduino Uno
for wireless communication between laptop (master) and embedded system (slave). Also interfaced potentiometer
(10 kΩ), VFD, and relay for complete automation chain.
◦ Dataset Generation & Pre-Processing for AI Applications: Created and labeled 800 RGB hand images
(same illumination, single location, low-res webcam). Applied image cropping as preprocessing and organized data
for binary classification using Google Teachable Machine.
1◦ Master-Slave Architecture for WiFi Systems: Implemented Bluetooth-based master-slave wireless
architecture with laptop as master/transmitter (computer vision + gesture recognition) and Arduino Uno as
slave/receiver (motor control execution). Achieved reliable indoor operation suitable for laboratory-scale industrial
automation.
◦ Technical Documentation for Engineered Projects: Authored complete technical documentation including
UML diagrams (use case, activity, sequence, class, deployment), wiring schematics, hardware specifications, testing
protocols (unit, integration, black-box, acceptance), and user guide.