Nicholas Howe

Software engineer and founder of CircuitCoder

Software engineer who builds products end to end, from microcontroller firmware to web and Android apps, and takes over unfinished codebases to get them to release. Founder of CircuitCoder, a Raleigh software studio for small and mid-sized companies.

Experience

Contract software engineer, Apex Instruments

Mar 2026 – present

XD-502 emissions-testing console (shipped)

Sole developer of the ESP32 firmware and built-in Wi-Fi web app for an EPA stack-emissions testing console, letting technicians monitor, control, calibrate and download tests from any phone or laptop.
Added over-the-air firmware updates, seven-language support and a two-tier password lock for calibration values.
Serialized commands from several simultaneous web users through one mutex-protected FreeRTOS queue.
Traced a Wi-Fi fallback fault to two root causes across the ESP32 and PIC32 boards and fixed both; made CSV export to SD card and USB drive reliable in the PIC32 firmware.

Next-generation automated sampling console (in development)

Leading a ground-up rewrite covering firmware for three boards (PIC32MZ, ESP32), a browser app and a Python hardware simulator: about 41,000 lines of C, warning-free under -Werror -Wconversion with no dynamic allocation.
Implemented EPA Method 1–5 isokinetic calculations in C, mirrored in Python and JavaScript with parity tests against the previous-generation console.

Founder, CircuitCoder

2026 – present

Ember, Android habit and accountability app

Streaks, XP, badges and 16 multi-week programs, all recalculated from versioned goal history so edits to past days stay consistent. Kotlin, Jetpack Compose, Room; 57 unit tests on the core logic.

Binder, Android trading-card scanner and price tracker

On-device OCR (CameraX, ML Kit) matched against four card-data APIs. Built an evaluation harness over 140 real card photos that raised top-1 match accuracy from 47% to 89% for Pokémon and 58% to 91% for Magic.

MicroPython starter guide, free electronics course

Hands-on lessons taking beginners from a first blinking LED on an ESP32 to a Wi-Fi-controlled robot.

Research

Neural networks for stock market prediction

Undergraduate research, UNC Asheville

Compared LSTM, CNN, CNN+LSTM and Transformer networks at predicting S&P 500 (SPY) moves from price history and technical indicators, using TensorFlow, scikit-learn and pandas.
[Key finding: which architecture performed best, and by how much]

Education

B.S. in Computer Science, University of North Carolina Asheville (transferred from NC State University)

2025

Skills

Languages: Python, JavaScript, C, C++, Java, Kotlin, PHP, HTML/CSS
Embedded: ESP32 (ESP-IDF, Arduino), PIC32 (XC32), FreeRTOS, over-the-air updates, Modbus TCP
Web and mobile: Android (Jetpack Compose, Room, CameraX, WorkManager), REST and WebSocket APIs
Data and ML: TensorFlow, scikit-learn, pandas, time-series modeling