I'm a professional nerd. That's the short version.

The longer version: I'm unashamedly curious and aggressively interested in how things work. I poke at things, I take them apart, and along the way I tend to spot problems and build something to fix them. Sometimes that something turns into a real tool people use. Sometimes it's an interesting demonstration of nothing in particular. Both are fine — that's how the learning happens.

If you scroll through my repos you'll find a mix of curiosity, failed ventures, and the occasional useful tool. No promises on the functionality of anything specifically. I'm here to be curious and learn out loud.

A note on what's happening here right now: I've been pulling old projects out of cold storage, blowing the dust off, and repackaging the parts that engineers around me keep asking for into clean, easy-to-deploy services. A few of those are below.

Khala — Agentic AI Teams My open R&D environment for learning the ins and outs of AI agents and agentic teams, built on the AWS Strands Agents SDK. Twenty named functional teams — strategy lab, nutritionist, sales, agent registry — coordinated through a planning and orchestration layer. Increasingly the place I spin up new agentic teams and run exploratory experiments.

Deepthought A graph-based reasoning engine that stores model weights as edges inside a Neo4j knowledge graph instead of dense parameter matrices. Learning is localized — only the edges connected to the tokens in a training example get updated, via Q-learning — which keeps the whole thing CPU-only and the reasoning paths inspectable. Named for the Douglas Adams computer; the bet is that systems that reason and explain themselves beat systems that only compute. Its lineage traces back to the RL core I built for Qanairy.

browser-service Hands out a real, dedicated browser instance over the network and keeps it alive for the session — open → interact → close, sticky on a session id, with a real-time socket and an HTTP fallback. Selenium 4 for desktop, Appium 8 for mobile. This is the clean spin-off of the Selenium browser farm that used to live inside Look-see and eat the GKE bill; instead of an embedded WebDriver in every consumer, it's a service that hands you a browser.

Unified-Data-Processing A Java 17 library that gives Kafka, Amazon MSK, Pulsar, Google Cloud Pub/Sub, and AWS Kinesis the same publisher/consumer interface — write the producer/consumer code once, swap brokers by swapping the wiring. On top of that sit DataBridge (fans heterogeneous sources into a unified Kafka backbone) and DataRelay (fans the backbone back out to any of the supported brokers), both at-least-once with provenance stamped on every message. It's the "stop writing the same ingestion shim for the fourth time" library.

Look-see — 2020–2024 An automated web accessibility audit platform that automates over 80% of WCAG requirements. Sole founder, sole engineer. MassChallenge 2021 cohort. The fun engineering bit was retiring two GKE clusters (a duplicated-monolith scale-out and the Selenium browser farm that ate the bill) and rebuilding on Cloud Run + Pub/Sub + Neo4j — operational cost went from ~$13K/mo to ~$640/mo. The browser-farm half of that work is what eventually became browser-service.

Using Semantic Versioning to Simplify Release Management — AWS DevOps & Developer Productivity Blog A release-management pattern for teams shipping fast without breaking downstream consumers.

Creating Your First CI/CD Pipeline Using GitHub Actions A practical, runnable walkthrough. 3,900+ reads and still compounding.

Mastering Concurrency with the Dining Philosopher Problem CS fundamentals treated as fundamentals, not trivia.

Navigating a Maze with the A* Algorithm in Python Pathfinding from first principles.

More at brandonkindred.medium.com.

Product for Product Management — EP 40: Startups, Look-See (Feb 2022) — with Matt Green & Moshe Mikanovsky. Founder story, accessibility audit product, what product managers should know about the lane.

Radio Entrepreneurs — Brandon Kindred, Qanairy, Inc. (Feb 2019) — with Nathan Gobes. Using AI to automate UI test generation and maintenance. YouTube cut.

Underserved — EP 029: Domo Arigato, Mr. Roboto (Nov 2020) — with Andrew Gelina. Path into programming, FIRST Robotics, AI/ML for software testing, founder lessons.

I've shipped production code in Perl, Kotlin, Python, Java, TypeScript/JavaScript, Scala, SQL, and Cypher. AWS (Lambda, API Gateway, EventBridge, Step Functions, ECS, DynamoDB), Terraform, CDK, Neo4j, Kafka. Problems over stacks — I pick up whatever the problem requires.

If you're looking for my professional work — clients, employers, the serious résumé version — that lives on LinkedIn. This page is the curiosity side.