Grant Drew
Cook on the Line, Rapid Technical Learner & Agentic AI Engineer
I've spent my working life in high-pressure operational environments — live broadcast audio, production lines, and for the past year a restaurant kitchen — and along the way I taught myself the technical work I wanted to do: QA testing, web development, Agile project management. Whatever I do, I give it my full effort. "Anything worth doing is worth doing well". Today, cooking is the day job: it pays the bills and carries the benefits. Indie game development — Godot, Bevy, and the wider Rust ecosystem — is the side hustle and the plan. The newest tool in the kit is a demonstrable agentic engineering practice: writing specs for AI coding agents like Claude Code and Gemini, directing multi-step autonomous builds, and reviewing every line they produce. This page is the proof — AI-generated under my direction, then human-reviewed, corrected, and extended.
Current Role Performance: Operational Cook (1 Year)
A year on the line. Below is the feedback as it was actually given — my manager's words in person, guests' words in public Google reviews. Checkable beats impressive.
What My Manager Said
He's proud of me, and he'd "recommend me." That's the whole quote — two things, plainly said. In a kitchen, that's how praise sounds.
- ✔ Standards held through peak-rush pressure
- ✔ Communication and backup across stations
What Guests Wrote
Guests have named me, and my restaurants, in the location's public Google reviews. Reviews I didn't write and can't edit — which is exactly what makes them worth citing.
- ✔ Order accuracy and food safety, ticket by ticket
- ✔ Consistent quality when the board is full
Functional Core Competencies
Instead of evaluating experience strictly by past job titles, this interactive matrix organizes candidate skills by functional capability. Click on a category below to explore specific technical tools, methodologies, and operational strengths.
Drive Claude Code and Gemini through spec-driven, multi-step autonomous builds — engineering the prompts and context, steering the work, and course-correcting in flight.
Select the right model for each job and chain them into one pipeline: Gemini generated this app from my spec; Claude Code reviewed and extended it under my direction.
Treat AI output as untrusted until verified: line-by-line human review, defect logging, and fixes — the same manual-QA rigor I applied to Android OS release testing.
Self-taught manual QA testing on physical Android devices and custom shipped OS variants, verifying system stability prior to release.
The side business in progress: game projects in Godot and Bevy, with side quests through the Rust web-server ecosystem.
Facilitated remote sprint cycles, led team video meetings, and managed Jira Kanban boards to streamline deliverables in a startup setting.
Acquired web development capabilities on the fly to maintain, update, and support company websites and web applications.
Kitchens, production lines, live broadcast: places where the work has to be right the first time, at speed, with no do-overs.
Maintained technical precision in live broadcast environments as Audio Tech (A1, A2) and Television Assistant (TVA) with zero room for error.
Strong attention to detail developed through strict cosmetic production packing standards (Lush Cosmetics) and exact recipe adherence.
Adaptability & Skill Transferability (Self-Assessed)
Full disclosure: these are self-assessments, visualized — my own read on where I'm strongest, not somebody else's measurements. The interactive part is real; the scores are one honest opinion.
Core Aptitude Radar Profile
Hover over the nodes for my self-assessed score in each domain.
Learning Velocity & Problem Solving
How quickly I've historically come up to speed in each domain — again, self-assessed.
Functional Experience & Track Record
Grouped by what I did rather than when I did it — kitchens to control rooms to codebases, the through-line is the same: learn fast, hold the standard, keep things moving.
Current Role: Line Operations & Culinary Execution
Active Position (1 Year)- • Manager's Endorsement: Told directly by my manager that he's proud of my work — and that he'd "recommend me." This page takes him up on that.
- • Guest Feedback: Guests have named me and my restaurants in the location's public Google reviews.
- • Teamwork: A busy line only works when every station covers the next — I communicate, jump stations when needed, and keep tickets moving.
Current Side Practice: Indie Game Development
Side Business in Progress- • The Gap, Owned: The stretch between tech jobs didn't take me out of software — it took me deeper in. Countless hours across Godot and Bevy game projects, laying the foundation for a business of my own.
- • Rust Ecosystem: Hands-on beyond the game engines — experiments across the Rust web-server ecosystem, from JWT-backed file servers to API prototypes.
- • Day Job, Side Hustle: Cooking pays the bills and carries the benefits; game development is the long game. Days on the line, nights in the engine.
Startup Technical & Agile Operations — Copperhead
Security-Focused Android Startup · Toronto- • Self-Taught Programming: Picked up programming and web frameworks on the fly to update and maintain company web applications and public-facing sites.
- • Operating System & Hardware QA: Conducted manual quality assurance testing on Android mobile hardware and custom OS variants (CopperheadOS), identifying stability issues prior to product deployment.
- • Agile Facilitation & Jira Management: Acted as Scrum Master, managing team Jira Kanban boards and leading remote video meetings to maintain task accountability and project momentum.
Precision Operations & High-Volume Logistics
Production & Technical Services- • Quality Assurance in Manufacturing (Lush Cosmetics): Maintained strict quality control, weight parameters, and visual standards in a rapid production environment.
- • Broadcast Audio Engineering (Audio Tech A1/A2 & TVA): Managed live audio mixing, hardware connections, and signal routing for television broadcasts and live events — where problems get fixed on air or not at all.
- • Self-Management: Kept multiple independent contracts and delivery commitments running at once — my own scheduling, my own accountability.
How This App Was Built: Directed AI Engineering
Yes — this application was built with AI, deliberately. I wrote the spec, directed Google Gemini to generate the first version, then ran a strict multi-model review: Claude Code plus my own line-by-line pass over the generated code. The review surfaced nine real defects — from a render-blocking CSS runtime, to a missing mobile menu, to inflated praise language in the copy itself — and all nine are fixed in the page you are reading, listed below as receipts. Using AI isn't the skill; directing it, reviewing it, and standing behind the result is. This page is where you check that claim.
Defined the structure, palette, interactions, and constraints, then briefed the AI with a precise, reviewable spec. The original directed spec is still visible in this page's source.
Google Gemini produced the full single-file application from that spec — layout, styling, charts, and interactivity — in one directed pass.
Multi-model QA: Claude Code assisted a line-by-line human review of the generated code — logic, styling, dependencies, performance, and metadata. Nine genuine defects surfaced — eight in the code, one in the words.
Nine findings, nine fixes — every one documented below. I own every line that ships, whoever (or whatever) wrote the first draft.
Real defects found in the AI's first draft during review, all fixed in this revision:
- ✔Fragile event handling: the tab filter relied on the deprecated implicit event global — refactored to pass the clicked element explicitly.
- ✔Asymmetric tab styling: the active-state logic removed classes that were never added and restyled the active tab as inactive — normalized so first paint matches every later state.
- ✔Grid display bug: filtered cards were re-shown with display:block inside a CSS grid — corrected to restore natural display.
- ✔Render-blocking CSS runtime: the Tailwind Play CDN compiled all CSS in the visitor's browser on every load — the single worst thing here for mobile. Replaced with static CSS precompiled once and inlined; instant first paint, no CSS network request at all.
- ✔Undeferred, unpinned Chart.js: loaded whatever "latest" a third-party CDN served, and blocked parsing — now version-pinned, self-hosted, and deferred.
- ✔No mobile navigation: the nav simply vanished below 768px — an accessible hamburger menu was added.
- ✔Duplicated comment: the generator echoed its "no SVG" confirmation twice — deduplicated.
- ✔Missing metadata: no <meta name="description"> or theme-color — added.
- ✔Inflated praise copy: the first draft dressed ambiguous input up as five-star "commendations" nobody actually gave ("flawless execution," "consistently commended") — caught on a later human pass and rewritten to claim only what's on record: my manager's own words, and public Google reviews.
Anyone can prompt an AI. The work is everything after the prompt: writing specs an agent can actually execute, picking the right model for each job, steering multi-step autonomous builds, and reviewing everything the machine produces — code and claims alike — with the same discipline I brought to manual Android OS testing.
- ✔The original AI spec still leads this page's source — view source to read it, followed by the revision note.
- ✔Gemini generated; Claude Code reviewed and extended — a working multi-model pipeline, not a claim.
- ✔Every agentic entry in the Skills Matrix above is backed by this build.
What Qualities Does Your Open Position Need?
Select what you are looking for in your ideal candidate. The tool highlights matching skills and direct evidence from my background.
Click a requirement to see the closest matching evidence from my background — plainly stated, nothing composite.
Awaiting selection...