About

From medical devices to
autonomous software — the long way around.

I started my career at PwC, then spent nearly a decade in medical-device sales. I was decent at it, but never loved it. I was also restless in a way I couldn't quite name — a sense that the work I was doing wasn't the work I was supposed to be doing.

On November 20, 2018 — at 34, on a whim and against my own skepticism — I opened Codecademy and started "Intro to Python." I didn't know what I was starting.

From there: Codecademy, Coursera, Udemy, then just building things. Self-taught into a full-stack developer. The pivot wasn't a plan; it was a compulsion. I kept making things because I couldn't stop being curious about things.

My fascination with AI started in 2017, after watching the AlphaGo documentary. Something about a system that could develop genuine intuition — not just pattern-matching, but something that looked like understanding — stuck with me. In 2019 I created the persona "AutonomousAJ" — a play on Alan Turing and John McCarthy, the father of AI. It was a way of naming what I was trying to become: someone who builds with intelligence as a collaborator, not just a tool.

The current method is what I call agentic development — running AI agents as genuine development partners. It's what lets a solo builder ship across as many domains as I do. Not about replacing human creativity; about amplifying it.

The curiosity domains — maps, food, aviation, sports, civics, history, quantum — aren't a portfolio strategy. They're just what I'm genuinely interested in. The pattern is that most of them end up becoming real, working software. That's not a plan either. It's just what happens when you're curious and you know how to build.

Andrew Pyle

Andrew Pyle

Full-stack · AI-native

Open to consulting

Technical range

PythonDjangoFastAPIPostgreSQLRedisCeleryTypeScriptReactNext.jsAWSDockerCI/CDNginxLinuxNetSuiteSuiteScriptERP Integration

AI tools — daily use

ClaudePrimary reasoning & code
Claude CodeAgentic dev partner
CursorIDE + AI pair programming
ChatGPTIdeation & drafts
GeminiLong-context analysis
GitHub CopilotInline completions
MidjourneyVisual generation
PerplexityResearch & fact-check
GroqLow-latency inference

Capability profile

Self-assessed, 0–100

A few dates

The long way around.

2017

The spark.

Watched the AlphaGo documentary. Something about a system developing genuine intuition — not pattern-matching, but something that looked like understanding — stuck.

AI fascination begins.

Nov 20, 2018

The pivot.

At 34, on a whim and against my own skepticism, I opened Codecademy and started "Intro to Python." I didn't know what I was starting.

Age 34. First line of Python.

2019

A name for the method.

Created the persona AutonomousAJ — a play on Alan Turing and John McCarthy, the father of AI. A way of naming what I was trying to become.

Alan Turing + John McCarthy.

Now

Building with agents.

Agentic development — running AI agents as genuine development partners. It's what lets a solo builder ship across as many domains as I do.

Currently exploring across every domain.

Build cadence
The compounding curve.
Cumulative showcased projects by first-repo year
The pattern
The first year: one small thing. Each year after: more. The domains compound — tooling built for one project accelerates the next. That's not a strategy. It's just what happens when you don't stop.
Years since the pivot
7
Nov 2018 → present
Projects
23
live + in progress
Active domains
8
built + exploring
Self-assessed capability profile
Honest about depth vs. breadth — Backend and Product are the deepest; DevOps is real but not the focus.
Merged-PR trend
How the shipping actually trends.
One daily-captured series — cumulative, rolling 12-month, and monthly throughput

Total merged PRs over time — it only climbs.