a passionate AI and Machine Learning undergraduate with strong programming skills in Python, C++, Java, and C. I have hands-on experience in machine learning techniques such as classification, regression, clustering, and computer vision.
Developing an Android malware detection system using static and dynamic analysis techniques. Using tools like Androguard and Wireshark to extract suspicious features from APK files and building an AI-based security reporting system.
Technologies: Python, Androguard, Wireshark, Machine Learning
Designed a deep learning pipeline for AI-generated speech detection, fine-tuned advanced models, and documented experiments with accuracy and loss tracking.
Technologies: Python, PyTorch, Deep Learning, ASVspoof2019 Dataset
Contributed to an open-source AQI analysis project. Created data visualization dashboards and enhanced forecasting models using machine learning.
Technologies: Python, Pandas, Matplotlib, Plotly, Machine Learning
I am currently interning as an AIML Developer at EPESIGN Technologies, where I am working on a state-of-the-art autonomous drone project. The core goal of the project is to enable a drone to navigate from Point A to Point B using onboard visual sensors (stereo cameras) and compute units like the Jetson Nano without relying on GPS.
As part of this, I am building a complete SLAM (Simultaneous Localization and Mapping) system using the VINS-Fusion closed-loop visual-inertial framework. This involves sensor calibration, ROS topic configuration, launching SLAM nodes, and visualizing pose estimation in RViz. The system is developed on ROS Melodic running on Ubuntu 18.04 (Bionic).
I work directly with stereo camera feeds and an Inertial Measurement Unit (IMU) to localize the drone and generate a real-time map of the environment.
Data fusion and trajectory optimization are handled within the vins_estimator module, while loop closure helps in drift reduction.
For flight control, I am integrating ArduPilot with a Cube Autopilot and using Mission Planner for simulation, calibration, and monitoring. Communication between Jetson and the Cube is achieved via the MAVLink protocol, which I configure using serial ports and ROS MAVROS bridge.
The setup involves multiple layers of development: launching ROS nodes, configuring parameters, verifying TF transforms, optimizing launch files, and integrating VIO into the control loop. I’ve gained a deep understanding of:
rqt_graph and rviz
🔗 Useful Resources Referenced:
– VINS-Fusion GitHub
– ROS Melodic Wiki
– ArduPilot Official Site
– MAVLink Protocol
– Mission Planner
This internship has greatly deepened my practical knowledge of embedded systems, drone software stacks, and real-world robotics development. I’m building both the perception and control pipeline for autonomous aerial vehicles.
Email: yashco.ltd@gmail.com
Phone: +91-9727835549
Location: Ahmedabad- Gujarat, India