An exploded drawing of the layers my projects work in. Choose a stage to see what happens in the circuit, the signal, and the software at that point.
Drawn for this site using a synthetic signal. Not a schematic of a specific board.
Founder and lead engineer, Mindset Technologies. March 2026 to now.
My work at Mindset connects neural-interface hardware, GPU signal processing, and software integration.
Read about the Mindset workAs an undergraduate researcher at Northeastern's Erdogmus Lab, my work focuses on low-cost EEG acquisition and embedded signal processing, separate from Mindset.
Custom EEG acquisition circuit boards designed in KiCad, with circuit review and simulation.
GPU signal-processing software written in CUDA and C++, with Python interfaces for working with EEG.
Python acquisition and classification workflows on Linux, and C and C++ interfaces for OpenXR, Unity, and Unreal.
Traces are synthetic and drawn for this site.
FPGA and embedded hardware, signal processing, and software tools. Each project page covers what I built, how it is put together, and the decisions behind it.
Show
Most of my signal-processing work is part of Mindset: the CUDA EEG library and the signal lab.
A local operations assistant that brings persistent memory, voice, coding agents, and engineering tools into one workspace.
Python, FastAPI, SQLite, React, TypeScript, Tailwind, Ollama
Read the case studyAn 8-bit ALU in SystemVerilog, integrated with a Kria KR260 and controlled from Python on Linux through AXI GPIO.
SystemVerilog, Vivado, Kria KR260, AXI GPIO, Python, Linux
Read the case studyA Next.js and TypeScript workspace connecting customer records, product configuration, quotes, and follow-up.
Next.js, React, TypeScript, Postgres, Drizzle, Clerk, Zod
Read the case studyA two-core RISC-V computer on the Kria KR260, with a small custom operating system and a fixed-point GPU in the FPGA fabric.
Kria KR260, Vivado, LiteX, VexRiscv, Verilog, SystemVerilog, C
Read the case study
Frame captured from the board's video memory.
GPU-pool job coordination for submitting model jobs, managing workers, and tracking execution.
Python, FastAPI, SQLite, Ollama, React, TypeScript, Swift
Read the case studyA TypeScript workflow for daily AI-agent briefings from structured logs, with email summaries, an issue-triggered phone path, and report history.
Built on OpenClaw agents and Inkbox's email, phone, and vault services, which are other teams' products.
A native desktop assistant that connects screen context with guided and agent-driven workflows.
Built with a teammate, who designed much of the notch interface and the run history and replay.
A creator-checkout demo with server-priced amounts, duplicate-submission handling, and a sales dashboard. Mock payments only.
An EMG signal-processing pipeline built with Holoscan and CUDA, connecting acquisition and replay input to processing and control interfaces.
A terminal coding assistant with local and remote model routing, tool use, and a focused command-line interface.
An extended OpenClaw administration console with sign-in and roles, safer configuration writes, and multi-bot workflows.
The original dashboard, its interface, and the Pixel Office view are xmanrui's work under the MIT licence. My changes are built on top of it.
A 6502 computer built from scratch, without a kit, with custom clocking, address decoding, memory banking, and memory-mapped I/O. My work included debugging with an oscilloscope and logic analyzer and programming in 6502 assembly.
My hardware internship at Aurus involved designing a solenoid-driven test fixture for payment-terminal inputs and writing Python automation for repeatable test cycles and measurement logging.
A Slack bot that pulls blockers and the people needed out of standup posts, emails them, replies in the thread, and follows up on old blockers.
A small Python experiment that plots a WAV recording as a waveform and a spectrogram with NumPy and Matplotlib.
I'm Nicholas Kozhemiakin, an electrical and computer engineering student at Northeastern University, graduating in April 2029. My work moves between FPGA and embedded systems, circuit boards, signal processing, and the software that ties them together.
Alongside leading engineering at Mindset Technologies, which I founded in March 2026, my work includes neural-interface research at Northeastern's Erdogmus Lab since January 2026. My sales and project-management role at Bang & Olufsen Boston involves audiovisual proposals and coordinating orders and installations. Earlier experience includes a hardware and robotics engineering internship at Aurus in June 2024.
My coursework includes embedded design, circuits, object-oriented design, and digital design, with brain-computer interface research supported by Northeastern's PEAK Base Camp Award in January 2026. Outside engineering, I build synthesizers and custom instruments, do woodworking, and make espresso.
Email is the best way to reach me.