Staff software engineer · Toronto, Canada
Hi, I'm Ash.
I help engineering teams use AI to ship faster without losing control of quality or cost.
Staff engineer in Toronto. Ten years at Silicon Valley companies. Today I run my own products with AI agents, and I help teams do the same.

Writing
Human-curated, typed with help from machines, and written to be useful to both.
How do I run my products with AI agents writing the code?
A case study from three days of building AI Agents Live 2.0, with a control plane in plain files, one gate for every change, a model measured against past human decisions and a hard cap on spend
- case study
- ai agent
- claude code
- verification
- llm cost
Which guardrails does an AI agent need before it goes to production?
A checklist of input, tool, output and operations guardrails to clear before real customers talk to your agent, with a pointer to the deeper guide for each
- ai agent
- guardrails
- checklist
- production
How to build guardrails for AI Agents
Put checks before the model, around every tool and after every reply, so your agent stays useful when a customer, a bug or a prompt injection pushes it off course
- ai agent
- rails
- security
- engineering
- tutorial
How do I test that my AI agent guardrails actually work?
Keep a small set of attack and edge-case prompts, run them on every change and watch real traffic, so a guardrail is proven and not just assumed
- ai agent
- guardrails
- testing
- evals
What does a custom guardrail for an AI agent look like in code?
Write one input check, one tool check and one output check as plain functions, then see where a framework helps and where it only adds weight
- ai agent
- guardrails
- code
- tutorial
How do I set up guardrails for an AI coding agent?
Limit what Claude Code, Codex or Cursor can read, run and push with permission rules, a sandbox, your own hooks and a review gate, so a wrong guess costs minutes and not your repository
- ai agent
- guardrails
- claude code
- codex
- cursor
How do I fix flaky Rails system tests when using Cuprite?
Most flaky browser tests come from timing, animations and shared state, and each one has a small fix that you can apply in Cuprite and Capybara today
- rails
- testing
- cuprite
- chrome
How to test in Chrome using Cuprite
Cuprite drives a real Chrome over the DevTools Protocol, so your Rails system tests are faster, steadier and closer to what users see.
- testing
- rails
- chrome
- cdp
How do I give an AI agent tools it can call safely?
Design each tool with a narrow job, validate what the model sends, ask a human before risky actions and limit what the tool can reach
- ai agent
- tools
- guardrails
- engineering
How to build your own AI Agents
Build a pantry agent that reads receipt and meal photos from a shared folder and leaves a shopping list on your phone every Friday morning
- ai agent
- ruby
- engineering
- tutorial
How do I structure marketing for an AI product?
Split it into three channels, the product itself, email and search, give each one an owner and a number, and judge every channel by purchases instead of clicks.
- marketing
- ai product
- geo
- seo
- strategy
Side notes
Older tutorials and things I made along the way.
How do I split a monolith into services when AI agents write the code?
People decide the boundaries and who owns the data, the spec becomes the thing you review, and each extraction stays small enough to undo
- architecture
- microservices
- ai agent
- migration
Why did I build a Chrome extension for rental hunting?
A rental hunt is a pipeline, and listing sites only give you a heart button. I built nivas, a Chrome extension for Kijiji, to fill that gap.
- chrome extension
- product
- rental hunting
- privacy
- sharing
How do I isolate tenants in Rails with PostgreSQL row-level security?
Bulletproofing multi-tenancy with a small migration helper, a tenant set once per request inside a transaction, and four traps to check before you trust it
- rails
- postgresql
- multi-tenancy
- security
How do I choose the right language model for each task?
Write the pass marks first, test every candidate on a fixed set of real cases, keep the model a config value behind one layer, and take the cheapest model that passes
- llm
- evaluation
- architecture
- ai engineering
How should I name REST resources, routes and associations?
Plurals for collections, a plural plus an id for one, and associations as plurals after the single. Naming conventions written so any stack can follow them
- rest
- api design
- naming
- architecture
How do I stop a RAG system from showing a user data they cannot see?
Put the asking user's permissions inside the retrieval query, draw the tenant boundary in the vector store, check again on every tool call and test it with leak cases
- rag
- ai security
- access control
- vector database
How do I get engineering teams to adopt an internal developer platform?
Make the platform road the easiest one, let teams serve themselves instead of filing tickets, mandate outcomes and not tools, and measure whether teams keep choosing it
- platform engineering
- developer experience
- leadership
- kubernetes
How to build an AI agent on ElevenLabs without writing a backend
A voice agent that answers questions, books meetings and lives on your website, built from the dashboard in an afternoon
- ai agent
- engineering
- product
How to leverage dynamic social image previews to delight your user and improve your GEO
Your users are sharing links, show them better content in image previews
- marketing
- engineering
- geo
- seo
How to market your AI Agent
You built your AI agent, congratulations! Now let's get it recommended when people ask AI what to use.
- marketing
- ai agent
- geo
- seo
How do I migrate many services to Kubernetes without downtime?
Move one service at a time behind a switch you can flip back, build the platform first, put the guardrails in the pipeline and judge the project by recovery time, pages and cost
- kubernetes
- migration
- platform engineering
- reliability
How do I reduce the on-call load on my engineering team?
Page only for symptoms that need a person now, give every alert one owner and a runbook, turn each incident into a change, and count the pages before and after
- on-call
- reliability
- incident response
- leadership
How do I lead a technical initiative across several teams?
Find a sponsor who owns the outcome, write the problem down before the solution, start with one team, treat a no as information, keep every group current and decide how it ends
- engineering leadership
- staff engineer
- technical strategy
- migration
How do I write API standards that every team in a large engineering organisation follows?
Write few rules, give each one its reason, check them with a linter in CI, keep people for the calls a tool cannot make, and give every exception an owner and an end date
- api design
- engineering leadership
- standards
- staff engineer
How should engineering team structure change as a software system grows?
Start with one team, then split by slice of the business, move the lines in the code with the lines between teams, add a thin platform when work repeats and watch for the signs you are behind
- team topologies
- conway's law
- engineering leadership
- platform engineering
When is a modular monolith better than microservices?
Draw and enforce the module boundaries first, then split out a service only when one module needs to be released, scaled or secured on its own
- software architecture
- modular monolith
- microservices
- domain-driven design
How can a tech lead coordinate several teams without becoming the bottleneck?
Move status, decisions and reviews out of your head and into writing, give each owner clear limits, and book your hands-on time before the meetings arrive
- tech lead
- leadership
- engineering management
- communication
How do I write a design doc that gets a decision?
Open with the decision, the people who make it and the date, then give the options with their costs, one recommendation, and a review that ends
- design docs
- engineering leadership
- architecture
- staff engineer
Which software design patterns are still worth using in everyday backend work?
Six patterns that keep earning their place, the ones I would drop, and a test for deciding whether a pattern pays or is only ceremony
- software architecture
- design patterns
- backend
- ruby
Things I'm building
AI Agents Live
Marketing and distribution for AI agents. List your agent, earn reviews and backlinks, and get structured for GEO so ChatGPT, Perplexity and friends can recommend it.
ShadowLink
A free Chrome extension that hides promoted posts, suggested content and promoted jobs on LinkedIn, so your Feed and Jobs pages show only what you chose. Private by design: no data collected, no network requests.
nivas
A Chrome extension for hunting rentals on Kijiji. Save listings, track each one from first look to lease, take notes, pin comments on photos and check the commute. Your data stays in your browser.