Archives

I let an AI agent work through my backlog over a weekend. I came back to 66 pull requests and still had to decide what was worth doing.

Fable 5 costs 2x Opus and burns like 6x. I spent a month learning to ration my wishes.

At 16 I passed off a Twilight Zone episode's plot as my own short film. My journal remembers what I did next.

I copied 27 Claude commands and realized it wasn't doing me any favors.

Waiting tables taught me something about managing multiple AI agents. Including when to stop taking on more.

I don't hand-write code anymore. The hard parts didn't go away.

Some of my best ideas used to die before anyone saw them. That's changing.

Working with AI isn't a binary skill. It's an instrument that requires deliberate practice.

Light Iron announced it was winding down. Some memories from my first job, my first family in LA, and where I accidentally started writing software.

Some feedback is easy to act on. Other feedback takes work to process. That work might be the point.

Working with AI changed what I'm willing to attempt. It also changed me.

Becoming a Staff Engineer meant working through self-doubt, helping others do better work, and learning which problems to leave alone.

Why do some conversations click while others fall flat? Using the language of networking to understand human connection, trust, and saying what you mean.

I built a sprint-planning game with AI over a weekend. Making it was more fun than playing it, and left me unsure what that means for the craft.

AI coding tools are everywhere, and each step feels reasonable. But what if we're all caught in an AI prisoner's dilemma, where individual choices lead to an uncertain future for our craft?

I started this blog for my future self. Eleven years later, writing has given me something I once felt I lacked: a record of things I made.

Choosing cinematography gave my film career direction. Moving into software later taught me that a clear goal can still leave room to change course.

I migrated my old Jekyll blog to Astro using AI tools. It cut the time in half, and I learned a lot about what AI is good at (repetitive tasks) and where it still needs human guidance.

Why a Python string comparison with 'is' appeared to work, and what string interning and bytecode revealed about identity versus equality.

An introduction to personal finance for engineers, covering budgeting, savings habits, investment plans, and the trade-offs between account types.

When I took over one-on-ones for a team of 8 engineers at Amazon, I needed a framework to make these meetings valuable. This template evolved over 20+ sessions to help create effective one-on-ones.

I had a tough career decision to make, so a manager suggested this simple quadrant exercise. It helped me map out the best and worst-case scenarios for both the short and long term, and it gave me clarity.

Feedback from mock technical interviews: tell specific stories, explain your contribution, know your language, and work with the interviewer's hints.

How I prepared for my 2022 senior engineer job search, from deciding what I wanted to practicing coding, system design, and behavioral interviews.

My promo journey at Amazon took two years from when I first thought I was ready. One rejection, a lot of reflection, and a manager's quote that made it all click.

Lessons from sending millions of production records through a new service: planning a backfill, testing small, monitoring, and verifying the results.

What four years at Amazon taught me about starting with customer problems, owning the outcome, and making time to keep learning.

Joining an experienced team can make every review comment feel presumptuous. How I get more comfortable offering feedback and asking questions.

Our service fell over at 20% of projected traffic in a load test. How I plan tests, define success, and use the results to find what needs fixing.

What I try when I'm stuck, how I explain the problem when asking for help, and what to do when nobody knows the answer yet.

From Starcraft maps to film school and post-production, the detours that led me back to software engineering.

A routine 45-minute wait became two minutes when I questioned it. Why staying sensitive to small frustrations can reveal worthwhile fixes.

A finished design can be hard to change. Showing the alternatives and decisions along the way gives reviewers a chance to help shape it.

I'm usually skeptical of certification programs, but AWS certifications felt different. They focus on how to build scalable and reliable services. My take on why they're worth it and how to prepare for the exams.

My study notes for the 2018 AWS Solutions Architect Associate exam, covering core services, practice tests, and how I prepared.

What I changed about code reviews after three more years: automate style checks, point out good work, offer context, and watch the tone.

I got into credit card churning a few years ago, and it's become a rewarding hobby. My wife and I went on our honeymoon using rewards points. This introductory guide covers the world of credit card rewards.

The blogs, books, talks, and courses that helped me grow as a developer, alongside a habit of regular practice and research.

A brief follow-up to the dangerous Python functions series, with an embedded exploit example for jsonpickle deserialization.

A trash pile, a tube of toothpaste, and two moving trains: three ways I think about improving legacy code and keeping rewrites manageable.

A step-by-step Flask setup using Vagrant and Ansible to automate a local server, with Gunicorn and nginx. Part one of a 2016 tutorial series.

Deploy the Flask app to Amazon EC2, then automate provisioning with Terraform and Ansible. Part two of a 2016 tutorial series.

Slides from my SoCal Python talk on improving legacy code, the risks of rewrites, and the two-trains analogy for switching systems.

I used to hate writing tests. Writing them as I go and treating them as first-class code made them useful enough to change my mind.

Lessons from my tech job searches and time as an interviewer, covering finding opportunities, preparing for interviews, and evaluating offers.

How I prepare a technical talk: write about the topic, choose a few takeaways, sketch slides, and rehearse aloud before the talk feels ready.

I found a list of Django security questions but couldn't find clear answers anywhere. So I dug in and wrote them myself, covering everything from CSRF and SQL injection to password hashing and mass assignment.

I felt like I didn't know enough as a developer, so I started listening to conference talks on my commute. Here are ten of my favorites that genuinely changed how I think about building software.

Slides from my SoCal Python talk on Bandit, including findings from scanning 16 open-source Python projects for potential security flaws.

A field report from testing Bandit on Python projects in 2015: running scans, interpreting warnings, and deciding which findings need attention.

The second installment covers pickle and YAML deserialization, SQL injection, and information leakage, with examples and safer alternatives.

I got banned from Ragnarok Online for "scamming" players with sunglasses. This bizarre experience taught me about cognitive biases and anchoring effects, drawing lessons from Daniel Kahneman's "Thinking, Fast and Slow."

Examples of command and code injection through Python's shell calls, exec, and eval, with safer alternatives for handling untrusted input.

Slides and resources from my LA Django talk on starting code reviews: what they offer, what to check, and how to discuss the code.

What I look for in a code review, from design and duplication to tests, and how I discuss changes with the person who wrote them.

I felt like a fraud when I became a full-time developer. Memories from learning to drive and working on film sets helped me put that feeling in perspective.

An experiment with UUID performance led me to submit a CPython patch. The benchmarks, proposed change, and steps I followed to contribute.

Notes from my commute listening: four Python and Django talks on service architecture, migrations, Unicode, and testable code.

Teaching my brother to program over weekend Skype calls taught me to keep projects simple, let him own the work, and make room for mistakes.

A short announcement of starting Georgia Tech's online computer science master's program, with Software Development Process as my first course.