I'm a systems engineer, organizational leader, musician, and a 2025 graduate of West Virginia University, focused on building scalable AI and robotics systems. I also founded and lead Deepiri, where I mentor developers from all over the world and build production scale automation tools and frameworks. Previously, I led the WVU Autonomous Racing Team to a 2nd place finish at the International F1Tenth Grand Prix and conducted robotics research in sensor fusion and autonomous navigation.
West Virginia University, B.S. in Computer Engineering - (Graduated May 2025)
F1TENTH Autonomous Racecar
Led the development of a 1/10th scale autonomous race car as part of WVU's F1TENTH team, achieving 2nd place at the 19th IEEE International F1TENTH Grand Prix. Engineered and tuned custom PID control loops for throttle and steering using VESC firmware, dramatically improving control precision and stability at high speeds. Implemented real-time path planning algorithms including pure pursuit, gap following, and dynamic obstacle avoidance using ROS2, Python, and LiDAR sensor fusion. Served as technical lead, learning ROS2 independently and training team members on framework integration, algorithm deployment, and hardware communication protocols.
Lydlr: Multimodal Edge AI Compression System
Currently designing a real-time multimodal sensor data compression system using ROS2 and PyTorch, targeting edge devices. At the moment, I'm deploying my nodes onto a Raspberry Pi, and eventually plan to utilize an NVIDIA Jetson. The system fuses camera, LiDAR, IMU, and audio inputs through convolutional and recurrent neural networks to achieve efficient learned compression with temporal context. Integrated a reinforcement learning-based adaptive compression controller that dynamically balances compression quality and system resource usage, guided by live perceptual quality assessment (LPIPS) and system metrics. I also prepared the data for quantization for AI edge hardware.
Aamati: AI-Powered MIDI Groove Synthesizer
Almost finished with my intelligent MIDI audio pipeline using the JUCE framework with C++ and PyTorch for real-time MIDI rhythm manipulation based on psychoacoustic modeling.
Keke: Real-time AI-Powered Excel Assistant
Built a native AI-driven FreeRTOS-based application for analyzing, cleaning, visualizing, and predicting from spreadsheet data. Developed the core processing engine, integrated intelligent data workflows with FreeRTOS, and designed an simple user interface to help with data analysis.
NeuralGPTOS
Designed and implemented NeuralGPTOS, an AI-native operating system with kernel-integrated agent runtime, eBPF-based semantic messaging, and vector memory services. Developed GPU memory management, sensory I/O drivers, and WASM/eBPF sandboxing to enable secure, high-performance AI agent execution directly within the OS.
ASM x86 Real Mode OS
Designed a bootable operating system written in pure x86 assembly, designed to run in real mode on legacy x86 architecture. Developed a custom 512-byte bootloader that initializes CPU, loads kernel from disk using BIOS INT 13h, and implements a command-line shell with keyboard input handling. Integrated FAT12 file system parsing, custom interrupt handlers through IVT modification, and basic memory management with segment:offset addressing. Successfully deployed on PlayStation 2 hardware using Free McBoot and a DMS3 modchip. Currently updating.
PortGremlin: USB Spoofing Firmware
Developed firmware for the EK-TM4C123GXL microcontroller that dynamically cycles through multiple USB device classes (HID keyboard, audio, printer, gamepad, MIDI) with randomized vendor/product IDs to spoof USB identities. Implemented timed USB descriptor switching and VID/PID spoofing using TI’s TivaWare USB library to test host OS device enumeration, driver stability, and security assumptions. Designed malformed descriptors to evaluate host resilience against unexpected USB device behaviors.
ExamJam: Interactive Full-Stack Study Platform
Developed a full-stack collaborative study platform specifically designed for West Virginia University students, providing free access to course-organized flashcards and study materials. Built using React.js frontend with Firebase authentication and MySQL backend, featuring dynamic search functionality, class-based organization, and responsive UI design. Implemented secure user authentication, real-time data synchronization, and intuitive course categorization to eliminate paywalls common in commercial study apps. Collaborated in an Agile development environment with Git version control, contributing to architectural design, component logic, and testing infrastructure.
404NotFound MPX OS
Designed and implemented a cooperative multitasking OS kernel using C and x86 assembly, mimicking early embedded RTOS architectures with precise timing and interrupt control. Developed custom interrupt service routines, system call interfaces, and process management using Process Control Blocks (PCBs) and Memory Control Blocks (MCBs). Implemented full context-switching mechanisms managing register states and stack frames, along with modular shell support, serial drivers (polling and interrupt modes), software timers, and real-time I/O scheduling without relying on existing operating systems.
MSP432 Microcontroller Robot
Programmed a fully autonomous robot using bare-metal C on the MSP432 microcontroller with no operating system. Integrated analog and digital sensors via GPIO and implemented real-time obstacle avoidance logic using timer interrupts and PWM-based motor control. Focused on efficient resource usage, low-latency decision-making, and precise motor speed modulation to achieve responsive navigation in dynamic environments. Demonstrated deep understanding of embedded systems design, peripheral configuration, and hardware-level programming.
VHDL RISC-Style CPU with Custom ISA
Engineered a custom 8-bit CPU from scratch using VHDL, emphasizing instruction-level control and finite state machine design. Created custom instruction set, hardcoded ALU, register file, and memory-mapped I/O interface with ModelSim simulation for timing analysis. Implemented FSMs for instruction fetch/decode/execute cycles, addressing real-world timing constraints including clock domain crossing and synchronous resets. Designed architecture supports basic arithmetic, logic operations, and I/O interfacing, laying groundwork for future FPGA-based projects.
AthletiDB
A sports data analysis and management system designed for sports analysts, researchers, and enthusiasts. This pipeline provides robust data collection, normalization, analysis, and export capabilities across multiple sports leagues.Discord Call of Duty Statistics Bot
Developed and deployed a feature-rich Discord bot for gaming community management, integrating with Call of Duty API (https://codapi.dev/) to track ranked play statistics. Implemented REST API communication, asynchronous programming patterns, and real-time message parsing for interactive user commands. Features modular event-driven architecture with comprehensive error handling, logging systems, and live testing protocols. Maintains active deployment with version control and community-driven feature updates.
UDP Probe Network Utility Tool
Built a comprehensive Linux-based command-line network diagnostic tool for UDP health assessment and latency analysis. Implemented custom socket programming in C with precise timing mechanisms for round-trip time calculation, jitter analysis, and packet loss statistics. Features robust logging modules and statistical visualization for network behavior analysis under variable conditions. Designed for network administrators requiring detailed UDP connection diagnostics and performance monitoring capabilities.
JaxK Full-Stack Website Template
Blank Full Stack Website Template for younger devs to make their own websites, as well as for me to host some of my projects.
Details
As I graudated from West Virginia University in 2025, I recieved an offer to work as an AI Software Engineer at Genaiva, a voice AI startup focused on saemlessly designing and integrating agentic restauraunt communication systems.
On top of my job, I serve as a Lead AI Systems Engineer at an R&D group that I founded named Deepiri, where I mentor aspiring software developers and teach them how to work on AI projects. I’ve designed digital circuits, built embedded robotic systems, and developed full-stack and AI-powered platforms from the ground up.
In 2024, I served as a team lead for the WVU Autonomous Racing Team, which competed at the 2024 International F1Tenth Grand Prix in Niagara Falls, Canada, where we secured 2nd Place.
I was also a Robotics Research Intern at the Intelligent Cyber-physical Systems Research Lab (iCPS Lab) at WVU, where I worked with analyzing sensor fusion data to develop efficient path planning and obstacle avoidance algorithms for mobile robots.
I've obtained SAS certifications in data analysis and statistical decision-making that I have utilized in my ML-based projects and automation pipelines. I'm always looking to grow through the challenges that come with any system I make, and can perform well wherever is needed in multiple hardware and software domains.
On top of my technical skills, I also worked outside for three of my college summers as a maintenance worker at my hometown cemetery, where I dug graves, pour cement, and wasn't afraid of hard work.
My entire goal is to create something that has everlasting impact.








