Portfolio
What I've actually built.
Some of these are tools I built for clients. Most are tools I built for myself, then kept using long enough to trust them. All of them started the same way, a real problem, not a demo.
Projects
AI system
dAvId Career Manager
An AI career assistant that turns any job posting into a suitability score, a gap analysis, and a tailored resume and cover letter, in the time it takes to make a coffee.
Tailoring a resume to every role is the single biggest time sink in a job search, and it multiplies for career counsellors doing that work for a full caseload by hand. I built a system that does the tailoring properly, every time.
What it does
- Pulls the company, title, and full job description from any public listing link, LinkedIn or a careers page
- Scores a candidate's fit out of 100 against their career history, with a gap analysis, delivered as a PDF in about a minute
- Generates a tailored resume and cover letter once a role passes, built from the strongest matching examples
- Manages the whole pipeline as a card-based board, New through Interview, so nothing gets lost
Under the hood
CLI tool
Indeed Apply Bot
A command-line tool that searches, scores, and shortlists Indeed listings for me, and never submits an application without me looking at it first.
Wading through hundreds of Indeed listings to find the handful worth your time is a slog, so I built a focused, scriptable tool for that one platform instead of a general-purpose application system.
What it does
- Scores every listing 0 to 100 against title relevance, skills overlap, experience match, and location fit
- Drops anything below the bar, plus hard exclusions like salary floor and specific companies
- Puts every shortlisted job on a review dashboard before anything gets submitted, human in the loop by design
- Logs every job it evaluates to a tracking sheet, so nothing gets tried twice
Under the hood
Web app
Post Writer
A tool that drafts my own LinkedIn writing without losing my voice to whatever an AI thinks “professional” sounds like.
Most AI writing tools flatten a specific voice into generic corporate copy the moment you ask for help, so I built one that starts from a written voice profile instead of a generic prompt.
What it does
- Suggests post topics grounded in live web search, not stale training data
- Drafts full posts against a structured voice profile, tone, vocabulary, sentence rhythm, and signature phrases
- Surfaces the sources behind every topic suggestion, so I can check a claim before I post it
Under the hood
Web app
DevTracker
A project and ticket tracker I built because I got tired of paying for tools that did more than I needed.
Most project management tools are built for teams of 20, with a learning curve to match. I wanted projects, tickets, and notes, in real time, and nothing else.
What it does
- Organises work into projects, each holding its own list of tickets
- Tracks type, status, and priority per ticket, with a running notes log so context never gets lost
- Syncs in real time, so changes show up without a refresh
- Signs in with Google, each user only sees their own work
Under the hood
Framework
The AI Transition Framework
A structured process for taking a business from “we should probably do something with AI” to a working, governed system, built around a proper development lifecycle instead of a demo.
Most AI pilots stall before production, not because the model fails, but because nobody defined the rules it has to follow or how a human stays in the loop. This framework makes sure that groundwork happens first.
What it does
- A discovery questionnaire that surfaces the real process, the real pain, and who owns the decision
- A self-serve ROI calculator with industry benchmarks pre-loaded, so a client with no data can still get a number
- A 9-slide briefing deck template a client's champion can take to their own leadership team
- A four-phase project plan template, with human oversight checkpoints built into each phase
Under the hood
Framework
Brand Voice System
A structured, reusable definition of a client's writing voice, built once, then used to keep every piece of AI-assisted content on-brand from that point on.
The usual approach to “write like us” is a one-line style instruction pasted into a prompt, and it shows. I wanted a voice definition detailed enough that any AI tool, or any writer, could pick it up and get it right without a re-brief every time.
What it does
- A structured voice profile covering tone, vocabulary, sentence rhythm, and platform-specific adaptations
- Real, approved samples of on-brand and off-brand writing, used to sanity-check the profile
- A machine-readable version of the same profile, ready to paste into any AI tool as a system prompt
Under the hood
Company site
1duck.tech
The site you're on right now, built and rebuilt in public as I figure out how to present One Duck Technologies.
A consulting site is the first proof point of the work itself. If I'm telling clients I can build clean, considered software, the site making that pitch needs to hold up to the same standard.
What it does
- Presents One Duck Technologies' services and portfolio to prospective SMB clients
- Statically exported and served directly, no server runtime required
- Built and iterated through a documented spec-and-plan process, every redesign has a written design doc before implementation
Under the hood
Let's go past the hype and answer the most important AI questions:Could we? Would we? Should we?
Book a callGet in touch
Tell me what you're working on.
Tell me a bit about what you're working on and I'll get back to you within one business day.
Prefer to talk it through?
Book a call