Engineering intelligent
software, robotics & avionics.
I'm Nolan Kurylo — a hands-on machine learning engineer working across avionics, autonomy and applied ML. At Valkyrie in Austin, I build the flight software and autonomy stack for an eVTOL R&D program — from wiring the hardware to writing the obstacle-avoidance control laws that run on it.
I'm a software engineer turned machine learning & avionics engineer who likes building systems that leave the lab and operate in the real world. I hold a Bachelor of Software Engineering from the University of Victoria with a specialization in Data Mining, Machine Learning & AI.
I'm a generalist by design. My career has deliberately crossed domains — production ML and data platforms, medical-device signal processing, semiconductor inspection systems, and now autonomous flight — because the most interesting problems live at the seams. I work end to end, from embedded firmware to deep-learning models and the cloud pipelines that feed them — and for fun I even brought up a bare STM32 flight controller as a personal project.
Originally from Calgary, I spent my early years playing high-level junior hockey — where I picked up the discipline and team-first mindset I still bring to engineering every day.
One engineer, many domains
I'm a deliberate generalist. Aircraft, medical devices, semiconductor equipment, energy data, e-commerce — every role has been a different domain and a different stack, and the pattern is the same: learn the problem deeply, then build whatever it needs.
Consulting-style engineering where every project is a new field. I drive hands-on avionics and autonomy on an eVTOL R&D program — an autopilot + Jetson Orin flight stack and a three-layer obstacle-avoidance system. I've been a senior engineer on production optical-inspection software for semiconductor manufacturing equipment — state machines, high-throughput capture pipelines and GPU-accelerated algorithms under hard timing tolerances. And in between I ship data and cloud systems: a Databricks change-data-capture ETL pipeline, a serverless Lambda that drafts weekly project status via the Claude API, and AWS governance standards for shared environments.
Energy-data platform work where I wore the ML, data-engineering and platform hats at once, helping lead a global team on a cloud-based, AI-powered web-scraping platform. Built an OCR table-detection model reaching 95% accuracy, engineered scraping optimizations delivering 25–50× speedups, designed an ML scheduling algorithm that cut cloud costs ~20%, and shipped NLP deduplication that reduced published table sizes 60% — with big-data processing on PySpark / Databricks.
A jump into medical devices: I taught myself biosignal processing from the research literature and developed a cuffless blood-pressure prediction model from photoplethysmography (PPG) signals, engineering beat-to-beat features into an RNN. Achieved MAE of 3.8 / 2.9 mmHg (systolic / diastolic) — meeting the AAMI and BHS standards for a "Medical Grade A" device.
Owned an ML system end to end at a small company — data acquisition, exploratory analysis, model design and deployment on AWS — building a content-classification system from a multi-input TensorFlow network with a fine-tuned BERT transformer, plus the DevOps to run it.
Where the range started: C++ firmware for PCB-based products — motors, sensors and accelerometers over UART/I2C with Bluetooth/WiFi — and a full-stack e-commerce platform built from scratch (Node.js, PostgreSQL, payments). Later led the web, mobile and embedded teams as a Software Engineering Team Lead, running weekly Kanban planning across all three.
What I build with
Avionics & Autonomy
- ArduPilot · PX4
- Pixhawk · Jetson Orin
- MAVLink · STM32
- Obstacle Avoidance
- Sensor Fusion
Machine Learning & AI
- Computer Vision
- LLMs · RAG · Claude API
- NLP
- Time-Series Forecasting
- PyTorch · TensorFlow
Cloud & Data
- AWS (Certified)
- Databricks · PySpark
- Lambda · CloudFormation
- ETL · CDC · Delta Lake
Languages
- Python
- C / C++
- JavaScript / TypeScript
- SQL
Featured projects
A few things I've built — from a from-scratch drone flight controller to computer-vision systems.
Autonomous eVTOL Flight Testbed
Hands-on avionics & autonomy on a work eVTOL R&D program: an autopilot + Jetson Orin flight stack and a three-layer obstacle-avoidance system with a physics-based velocity governor.
Custom STM32 Flight Controller
A quadcopter built from the flight controller up — a bare STM32H743 running ArduPilot with a hand-authored hwdef, bringing up IMU, baro, radio and ESCs myself.
Duck Hunt AI
A computer-vision agent that plays Duck Hunt, fine-tuning YOLOv5 and SSD detectors via transfer learning for high mAP at real-time speeds.
Let's build what flies next.
I'm exploring new roles in machine learning, autonomy and aviation. If you're working on something ambitious in the air, I'd love to talk.