I'm a marketing and systems manager for B2B and B2G software companies. I build the systems that capture and track every lead, create AI assistants that watch ad spend so people don't have to, and make brands the answer buyers get when they ask ChatGPT, Perplexity, or Google for a recommendation.
Most marketers pick a lane: strategy, systems, or execution. I run all three, and I use AI to run them at a scale one person shouldn't be able to reach.
I set up HubSpot, the system that stores every lead and customer, from scratch: scoring which leads matter most, sending each one to the right person, and building dashboards that show leadership exactly which euro produced which deal.
AI assistants that check ad accounts every morning, tools that answer data questions in plain English, and content strategies that get brands recommended by name inside AI tools like ChatGPT.
Paid advertising across Google, LinkedIn, Meta, and Microsoft. Deep research into ideal customers, turned into messaging, landing pages, and structured testing that cuts the cost of every lead.
Every project below is real, in production, and built end to end by me. Open any card for the full build: architecture, tooling, and results.
Built with n8n, the Claude API, GitHub, and VS Code. Every morning at 9:00 AM it flags problems, sums up performance, and suggests stronger ad ideas across Google, Meta, and Microsoft Ads, delivered straight into Slack.
A content strategy built so AI tools recommend the brand by name. Reached a 48% citation score and a 65% sentiment score, making the brand the most cited in its category across AI search.
A connector I built that links website and search data directly to an AI assistant. Any question about traffic, search terms, or pages gets answered in seconds, just by asking.
15 forms, 94% unknown attribution, no forecasting. I rebuilt tracking, ranked every form by conversion, and gave leadership a pipeline they could finally predict.
Three connected automations: every new blog post promoted on LinkedIn automatically, every LinkedIn lead traced back to the exact ad that produced it, and every Reddit mention of the brand flagged in Slack in real time.
Conditional routing that sends every query to the right team, plus end-to-end webinar automation from registration to Zoom to reminder email. No inbox chaos, no manual steps.
Led a 4-person team through a 4-month rebuild: 15+ lead magnets, intent-based pop-ups, a qualifying chatbot, and full copy across 18 pages. Launched at a 92% SEO score.
A Clay.com pipeline that automatically adds company details to every new lead, scores how well each one matches the ideal customer, and routes them into the right HubSpot follow-up automatically.
Structured campaign execution across Google, LinkedIn, and Microsoft Ads with data-led budget allocation and an A/B framework that lifted conversion rates by 15 to 20%.
Using a chatbot is table stakes. The projects below are AI doing real, scheduled jobs: an assistant that reviews ad accounts every morning, a tool that answers data questions instantly, and content built to be the answer AI gives.
An assistant built with n8n and Claude that reads the Google, Meta, and Microsoft Ads accounts every morning, then posts a clear report to Slack at 9:00 AM: problems flagged, performance summarised, better ad ideas suggested.
I built a custom connector so analytics stops being a dashboard chore. Any question about website traffic or Google search performance, asked in plain English, answered from live data in seconds.
AI tools answer questions directly, and if your brand isn't named in the answer, you don't exist. I restructured the website's content so AI systems can easily read, trust, and quote it, and tracked the results in HubSpot.
Everything here has been used to build something in this portfolio.
Based in Dublin, open to relocating to Canada. If you need someone who can architect the CRM, deploy the AI agents, write the copy, and defend the numbers, I'm one email away.
I built an AI agent that does the first 90 minutes of a paid media manager's day without the paid media manager. It catches campaign issues, summarises performance, and recommends better ad angles, delivered to Slack every morning at 9:00 AM.
Paid media performance degrades quietly. A broad match keyword starts draining budget on a Tuesday night, a winning creative fatigues over a weekend, a tracking pixel breaks silently. By the time a human reviews the account, the waste has already compounded.
Reviewing three ad platforms properly takes over an hour a day. Doing it inconsistently is worse than not doing it at all, because it creates false confidence. I wanted the review to happen every single day, at the same depth, whether or not a human showed up.
An agent orchestrated in n8n that runs on a daily schedule. It pulls fresh performance data directly from Google Ads, Meta Ads, and Microsoft Ads, hands it to Claude (Anthropic's AI) with detailed instructions on how to analyse an ad account, and posts the finished report to a Slack channel at 9:00 AM.
Everything is stored in GitHub and built in VS Code, so every change to the logic is tracked and reversible, the same discipline used for real software.
When buyers ask ChatGPT, Perplexity, or Google AI Overviews for software recommendations, the brands cited in the answer win. I built a content strategy engineered for LLM citation and made a B2G SaaS platform the most cited brand in the grant management category.
Traditional SEO assumes the buyer clicks a blue link. Increasingly they don't. They ask an AI engine, read the inline answer, and shortlist whichever brands the model cites. Rankings you spent years building don't transfer automatically, citation behaves by different rules than ranking.
The category was wide open: no competitor was deliberately optimising for answer engines. That was the opportunity.
Vanity claims are easy in AI search, so I measured it properly. Using HubSpot's AI visibility tracking, I monitored two scores continuously: citation score (how often the brand appears in AI answers for category queries) and sentiment score (how favourably it's described when it does).
Dashboards only answer the questions they were set up for. I built a custom connector (using MCP, an open standard for linking AI to live data) so I can ask Claude anything about website traffic or Google search performance and get an answer from live data, in seconds.
Real analytics questions rarely match a pre-built dashboard. "Which pages lost clicks after the last core update?" means exporting from Search Console, filtering by date ranges, cross-referencing with GA4 sessions, and assembling the story by hand. Twenty minutes of work for a question that should take twenty seconds.
The result is that questions simply don't get asked. Data-driven marketing dies not from lack of data, but from friction in reaching it.
A connector running on Google Cloud, securely linked to Google Analytics 4 and Google Search Console. Once plugged into Claude, both data sources become something you can simply ask questions about.
Fifteen forms, thousands of views, and 94% of contacts with unknown attribution. I rebuilt tracking from the ground up, ranked every conversion path, and gave leadership a pipeline they could forecast instead of guess.
Blog posts sat unpromoted, LinkedIn conversions went untracked, Reddit mentions were missed. I designed three interconnected automations that turned a B2G SaaS marketing operation into an always-on system.
Every form submission hit everyone's inbox. Webinars ran on manual effort at every step. Nobody knew which product categories drove demand. I fixed all three with two interconnected HubSpot systems.
A conditional workflow that evaluates every incoming query and routes it to the specialist who should answer first:
No more inbox clutter, no more three people answering the same email, and measurably faster first responses.
What used to require manual intervention at every step now runs itself: a matching form submission triggers automatic Zoom webinar registration, a processing delay for system sync, then a personalised confirmation email with the join link, timed 24 hours before the event.
I restructured form submissions to capture product category interest, giving leadership real-time visibility into which offerings pull the most demo and quote requests, demand data that now informs product and campaign strategy.
A B2B SaaS website with no lead capture, weak SEO, and no conversion paths. I led the complete transformation as project lead: strategy, all the copy, technical SEO, and a 4-person team over 4 months.
Manual lead research is slow, inconsistent, and unscalable. I built a Clay.com pipeline that enriches every inbound contact with firmographic and technographic data, scores ICP fit automatically, and routes leads into the right HubSpot lifecycle.
Structured campaign execution across Google, LinkedIn, and Microsoft Ads, grounded in ICP research and disciplined A/B testing, with a full HubSpot CRM architecture underneath to score, nurture, and report on everything the ads produced.
Ads without infrastructure just buy expensive anonymous traffic. Before scaling spend, I architected the HubSpot CRM and Marketing Hub from scratch: lead scoring, automated nurture workflows, and segmentation across 5,000+ contacts, the company's first scalable marketing ops layer. A revenue pipeline dashboard tracked MQL-to-SQL conversion so leadership could see marketing-attributed pipeline in real time.