I build the ops layer that makes Marketing and Sales teams successful at scale.

Problem

A hospitality group running large-scale corporate events was managing attendee check-in with printed lists and manual tick-offs. Entry queues backed up badly at peak arrival windows, staff couldn't see real-time attendance counts across multiple gates, and post-event data reconciliation took hours.

Solution

Built a web-based QR code event management system: unique QR codes generated and emailed per registration, a mobile-friendly scan interface deployable on any device without app installs, multi-gate support with real-time sync, a live attendance dashboard for event ops, and automatic post-event CSV export.

Problem

Retail investors in a frontier market had no reliable way to surface signal across 277 listed equities. Finding meaningful moves meant manually checking multiple public portals, open data sources, and broker email drops each morning, a process that took hours and still missed things.

Solution

Built a full-stack market intelligence platform: live ingestion from public and open-source market data, a portfolio tracker with cost-basis P&L, AI-generated morning briefs via WhatsApp before market open, email alerts on research report drops, and a web dashboard for historical analysis.

Problem

A high-growth AI company was running 40+ field events annually (conferences, executive dinners, partner booths across North America and Europe). After each event, the ops team had to manually reconcile attendee lists, apply campaign statuses in Salesforce, trigger nurture enrollments, tag prospects in Outreach, and notify SDRs. A single mid-size event took 25 to 30 hours of ops time. By the time SDRs saw the data, it was two days stale.

Solution

Rebuilt the entire pipeline as an orchestrated, near-zero-touch system. A multi-stage n8n workflow ingests the event sheet, maps attendance statuses, upserts contacts into HubSpot, writes campaign member records in Salesforce, fires Outreach tagging webhooks, and delivers channel-specific Slack summaries to each audience.

Problem

Inbound leads were reaching the wrong reps, in the wrong order, on the wrong day. A complex mix of territory rules, persona signals, round-robin pools, and CRM ownership logic had accumulated across two systems with no single source of truth. High-intent leads went cold. Revenue attribution was unreliable.

Solution

Designed and implemented a multi-tier intelligent routing system spanning HubSpot and Salesforce. Routing logic accounts for persona classification, company signals, territory assignments, rep capacity, and ownership hierarchy, without exposing the underlying rules to downstream systems.

Problem

The asset management arm of one of Canada's largest financial institutions handled inbound interest across four product lines (wealth management, fixed income, equity funds, and structured products) with a flat point-based scoring model that treated a high-net-worth institutional referral identically to an early-stage retail enquiry. Advisors had no prioritisation signal, no visibility into what drove a score, and no way to distinguish a ready-to-engage lead from noise. Conversion from scored inbound to a first advisory meeting sat below 20%.

Solution

Built a multi-track lead intelligence platform running four parallel scoring dimensions simultaneously: a behavioural track (web depth, document downloads, calculator interactions, email velocity); a profile-fit track (entity type, estimated AUM range, jurisdiction, declared vs. browsed product affinity); an intent track (recency-weighted activity scores with exponential time-decay applied per signal class); and a relationship track (existing account depth, advisor contact history, referral chain, org linkages). Each dimension produces a normalised sub-score. The composite score feeds into segment classification and advisor queue prioritisation in under 90 seconds. Every component is logged with its contributing weight and triggering signal, a hard requirement given the regulatory environment.

Problem

Running complex multi-domain operations across marketing, data, finance, and product without a coordination layer meant constant context-switching, inconsistent outputs, no persistent memory between sessions, and brittle manual handoffs between tools.

Solution

Built a web-based AI coordination system: a multi-agent application with a dashboard, domain-scoped knowledge bases, automated task routing, and quality gates before consequential actions execute. Handles recurring workflows across six domains with minimal manual intervention.

Problem

The GTM team was spending around 10 hours a week manually scanning competitor sites, review platforms, and job boards for pricing changes and positioning shifts. The output was inconsistent, always late, and rarely made it into a decision anyone could act on.

Solution

Built a scheduled intelligence system that scrapes 30+ sources, runs LLM-powered signal extraction, flags meaningful changes, and delivers a structured Slack digest every morning. The team reviews a single ranked summary instead of doing the research themselves.

Problem

Every visitor to the site saw the same homepage regardless of industry, traffic source, or intent. The team was running A/B tests that took months to reach significance and never addressed the core issue: the content was too generic for any one segment to feel spoken to.

Solution

Led implementation of a rules-based dynamic content system that swaps hero copy, value proposition language, social proof, and CTA based on traffic source, referrer, and form enrichment data. No engineering dependency for new segments: the marketing team owns the rules.

Problem

A fintech startup was delivering portfolio data to clients through an API that was slow, occasionally wrong, and had no audit trail. Clients were catching errors before the team did. There was no way to demonstrate data provenance to regulators or enterprise buyers who asked.

Solution

Specified and led implementation of a validated data delivery pipeline with Salesforce and SFMC at its core: input validation at ingestion, checksums at delivery, structured error logging, and an append-only audit log per record. Deliveries went out via personalised email; each client received only data relevant to their portfolio, not a generic report.

Problem

A support team handling 400+ monthly tickets was responding in 36 hours on average, mostly to the same recurring questions. Separately, inbound demo requests were sitting for a day or more before a rep picked them up, and the warmest leads had gone cold by then.

Solution

Configured an AI support flow connected to the knowledge base that handles tier-1 queries and escalates with full context. Added a qualification bot for inbound demos that collects use-case details, scores fit, and books calendar time — so reps only touch leads that are ready.

Problem

A high-velocity B2B sales team was building prospect lists manually using static databases that couldn't capture real-time market signals. Funding rounds, executive changes, technology migrations, and expansion announcements (the moments when a company is most likely to buy) went undetected until they surfaced in a newsletter days later. By the time a rep reached out, competitors had already booked the first meeting. Prospecting consumed over 12 hours per rep per week with no systematic way to prioritise who to contact first.

Solution

Built an autonomous company scouting engine that continuously monitors 12+ signal sources in parallel: news APIs, LinkedIn company activity, regulatory and funding databases, job board postings, and technology fingerprinting from public signals. A custom entity resolution layer clusters multiple signals about the same company before scoring begins, preventing duplicate noise and ensuring a funding announcement, three new engineering hires, and a product launch press release are all understood as a single high-intent company moment. A trained AI model — loaded with ICP criteria and domain knowledge bases, evaluates each resolved signal cluster and generates a brief rationale for every match. High-confidence prospects are automatically enriched and staged in the CRM with full signal provenance attached. Reps receive a ranked daily digest, no research required.

Problem

A B2B SaaS company with hundreds of addressable use cases, integrations, and target verticals needed search presence across long-tail queries without scaling a content team proportionally. Hand-written pages took three to four weeks per batch, couldn't keep pace with product changes, and the backlog grew faster than it was cleared. Competitors were ranking on queries where the product clearly had the better answer, simply because they had content and the company didn't.

Solution

Built a programmatic content engine: a structured schema maps every relevant combination of use case, integration, vertical, and feature. A trained AI model configured with brand guidelines, product knowledge bases, and internal linking rules generates page content from that schema. A validation pipeline checks each generation for factual alignment with the product database and brand guideline compliance before publishing. Pages below a confidence threshold are routed to a human editor with a diff view of what the AI produced vs. what the validator flagged. Approved pages are published directly to the CMS, submitted to the search index, and connected to the analytics stack. A change-detection layer refreshes pages automatically when underlying product data is updated.

Problem

Businesses across consulting, finance, hospitality, and professional services were running on outdated sites, WordPress templates they couldn't modify, or no web presence at all. For international clients the problem was positioning — sites that looked generic in a market where first impressions determine whether a prospect reads further or bounces.

Solution

Designed and built custom sites for individual professionals, small businesses, and growing companies, from personal brand sites and portfolio pages to full product landing pages and event microsites. Each project started with a positioning conversation before any design decisions. Built on modern stacks for performance and maintainability, with clear handoffs so clients could manage content without touching code.