Passionate about exploring new technologies, with a profound love for robotics. Dedicated to integrating embedded systems with artificial intelligence, aiming to create innovative solutions that advance the field and push the boundaries of what is possible.
Download CV
A fervent devotee of modern technology, I am driven by curiosity and a constant desire to learn,
with a strong interest in building intelligent systems that bridge software and hardware. My
work often lies at the intersection of artificial intelligence, embedded platforms, and
robotics, where ideas are transformed into practical, real-world solutions. I enjoy working with
modern AI paradigms, from advanced language models and intelligent agents to data-driven
reasoning systems, while also exploring vision-based understanding and model architecture
design. On the hardware side, I have hands-on experience with embedded platforms and low-level
system design, enabling me to create solutions that are both intelligent and efficient.
With a strong programming foundation and an appreciation for system-level thinking, I
focus on developing scalable, reliable, and thoughtfully engineered solutions. I actively keep
pace with evolving technologies and enjoy experimenting with new approaches that challenge
conventional boundaries. Motivated by innovation and problem-solving, I aim to contribute
meaningful value while continuously refining my understanding of intelligent systems and their
real-world impact.
Hobbies
Qualities
Education
Languages
Achievements
C++
Python
Embedded C
CUDA
VHDL
Verilog
STM32 CubeIDE
NVIDIA Jetson Nano
Raspberry Pi
Arduino UNO / MEGA
Large Language Models (LLMs)
TinyML
NanoEdge AI
Prompt Engineering
OpenCV
OpenGL
LangChain
LangGraph
AMD Vivado
RTL Design (VHDL / Verilog)
Altium Designer
Cadence
TCAD / Microwind
Multisim
MATLAB
GitHub (Version Control & Collaboration)
A vision-driven robotic system that interprets human hand gestures using computer vision and control logic, enabling intuitive human–robot interaction. The bot processes gesture inputs in real time and translates them into precise motion commands for responsive navigation and control.
A smart IoT-enabled emergency response system triggered via an in-vehicle push button. Upon activation, it broadcasts real-time alerts to nearby vehicles and traffic authorities, enabling adaptive traffic prioritization and reducing emergency response latency through coordinated traffic management.
A cloud-assisted drone architecture that offloads computationally intensive vision tasks to a remote server via 5G/Wi-Fi connectivity. Images captured by an onboard Raspberry Pi are transmitted for YOLOv8-based inference, significantly reducing onboard power consumption while extending flight endurance and operational range.
A hybrid deep learning framework combining Swin Transformer and ResNet50 architectures to improve pneumonia detection from medical images. The approach leverages transformer-based global feature modeling alongside CNN-based spatial representations for enhanced diagnostic accuracy and efficiency.
An embedded health monitoring system built on STM32, integrating MAX30102 and MLX90614 sensors to acquire biometric parameters such as heart rate, SpO₂, and body temperature. The collected data is processed for early-stage disease prediction, enabling efficient and low-power edge-level diagnostics.
An embedded diagnostic system that integrates a digital stethoscope with an STM32 microcontroller to analyze lung sounds. MFCC-based feature extraction and efficient signal processing are employed to enable accurate pneumonia prediction on resource-constrained hardware.
A custom-designed flight computer for rocketry applications featuring an STM32F411RE-based PCB. The system integrates GPS, IMU, temperature, and humidity sensors, enabling real-time data acquisition, processing, and wireless telemetry via an nRF24L01+ module for precise navigation and environmental monitoring.
This project designs a custom 19-bit CPU architecture using Verilog HDL, featuring a specialized instruction set for signal processing and cryptography. The CPU implements a 5-stage pipeline, a 16-register file, and supports both standard and custom operations like FFT, encryption, and decryption. The lightweight, efficient design makes it ideal for embedded domain-specific applications requiring high-performance processing.
In emergency scenarios, a smart IoT-enabled system is activated via a strategically placed button inside emergency vehicles. Upon activation, the system instantly broadcasts synchronized alerts to nearby vehicles and traffic control authorities, enabling dynamic traffic prioritization and optimized signal management to ensure rapid and unobstructed emergency response.
Learn moreDeveloped an edge–cloud drone vision system that reduces onboard power consumption by offloading AI processing to the cloud. A Raspberry Pi 5 mounted on the drone captures images at fixed intervals and transmits them over high-speed 5G Wi-Fi for server-side analysis using a YOLOv8-L object detection model. This architecture significantly extends flight time while enabling real-time human and vehicle detection, demonstrating an efficient and scalable approach to energy-aware aerial surveillance.
Learn moreDesigned and orchestrated advanced AI systems leveraging prompt engineering, diffusion models, multi-model integration, rigorous training, validation, and performance optimization pipelines.
Learned large language models, tokenization strategies, and engineered agentic workflows enabling autonomous reasoning, tool orchestration, context management, and scalable decision-making.
Led end-to-end embedded AI development, integrating biomedical sensors (MAX30102, MLX90614) and optimizing low-latency algorithms for real-time, resource-constrained processing.