Most marketing consultants stop at strategy decks and templated dashboards. I keep going.
What follows is a working tour of the systems I've built — for clients, as products, and as research. Custom dashboards that replace $7K–$21K/yr of off-the-shelf tooling. Multi-platform data pipelines across every major ad network. Full-stack SaaS products with auth, billing, and AI. And the marketing engineering glue that holds it all together.
If you're hiring a marketing operator who can't ship code, you're locking yourself into someone else's product roadmap forever. I'm the alternative.
Production integrations
Scale of work in flight today
Live systems, datasets actively maintained, and products with paying users — not aspirational case studies.
Four pillars
The work falls into four categories. Most engagements cross more than one — a dashboard isn't useful without the data pipeline behind it; a SaaS isn't useful without the marketing tooling around it.
Custom client dashboards
Branded, live, yours.
Hash-routed multi-tab dashboards pulling from any marketing API. Real-time refresh. Per-client branding tuned to each customer. PDF and CSV export. Password-gated access. Drilldown from account to campaign to ad group to individual asset performance.
Marketing data plumbing
The unsexy work that makes everything else possible.
OAuth flows across six providers. Token refresh logic. Multi-account merges that roll nine ad accounts into one brand/non-brand view. BigQuery joins between GA4 events and Google Ads conversion actions. Conversion-action segmentation. Custom UTM frameworks with dynamic ad-group slug injection.
Full-stack SaaS products
Multi-tenant, billed, secured.
Next.js plus Supabase Auth plus Stripe checkout plus webhook reconciliation plus row-level security plus tier-gated feature flags. Alerting engines, watchlists, API key management, email and SMS notifications. Built so a real human can sign up, pay, and use it — not just a demo.
AI-enabled tooling
Useful Claude and GPT integrations — not chatbots.
Natural-language search that compiles to SQL. Intent classification routing prompts to the right tool. Content-generation pipelines with compliance scanning that blocks the model from introducing violations. Vector embeddings plus pgvector for semantic similarity. Multi-model dispatch with model-specific quirks handled.
The fastest way to be locked into your competitor's product roadmap is to build your stack out of other people's products. The fastest way to outflank them is to build your own.
Systems running for real businesses
Each of these is delivering data to a paying client today. Specific client names are kept private; described here by industry and shape of the engagement.
Competitive intel suite + live performance reporting
Two-part build delivered as a unified client portal.
A keyword competitive-intelligence platform scraped and enriched tens of thousands of keywords across six direct competitors, segmented by brand, topic, and intent. Alongside it, a six-tab live performance dashboard with vendor-versus-affiliate signup attribution that joins the client's GA4 BigQuery export with Google Ads per-conversion-action data — closing an attribution gap that GA4 alone can't.
Financial operations + multi-source ops layer
High-stakes engagement supporting a court-overseen restructuring.
Six years of margin reports, ad spend, bank statements, debts, and forecast scenarios merged into a financial dashboard used in court filings. Alongside it, a live ops layer: tens of thousands of call records ingested through a custom CSV pipeline, ActiveCampaign CRM integration that creates deals on missed calls via Outlook → Zapier email-parser → custom webhook, and an Airtable rebates sync exposing six-figure outstanding payments.
Cross-platform reporting + SEO audit + Squarespace asset
Read-only ad performance across four platforms, on a single screen.
A multi-platform dashboard merging Google Ads, three separate Meta accounts unified into one view, TikTok Ads, and Google Search Console. Warm-gold and cream design matching the gallery's print aesthetic. Plus a standalone local-SEO audit diagnosing Map Pack absence, and an Instagram-reviews snippet drop-in for the client's Squarespace site.
Multi-platform paid media + expert-witness production
Two engagements in one — paid media reporting and litigation deliverables.
A reporting layer pulling Google Ads, Meta, and Microsoft Advertising. The Bing integration required a hand-rolled SOAP XML client because Microsoft's REST endpoints don't reliably work for the operations needed. Plus Google Business Profile location export and an expert-witness production line — rebuttal reports, PDF generation, exhibit packaging via Chrome headless rendering.
Multi-account portfolio rollup
One dashboard, nine Google Ads accounts, unified across the portfolio.
Single-pane reporting across nine separate Google Ads accounts spanning the holding company's brands. Custom brand-versus-non-brand classification with priority logic. Monthly trend aggregation. Per-account conversion-action breakdowns. Linkable tab routing so the executive team can share precise drilldowns by URL.
Non-brand
Pattern-source dashboard
The first build of this generation — pattern source for everything that came after.
Single Google Ads account, five tabs covering overview, campaigns, ad creative with per-asset performance labels (Best / Good / Learning / Low), search terms, and keywords with quality-score visibility. Asset-level performance surfacing isn't typically available in templated dashboards; here it's the centerpiece.
SaaS I've shipped on my own
Each of these is publicly accessible — designed, built, deployed, and operated end-to-end. They're the proof that the same systems thinking that delivers your client work also ships real businesses.
SportsBookish
Pricing-edge analytics across Kalshi versus traditional sportsbooks.
A three-tier SaaS surfacing arbitrage between Kalshi event contracts and the sportsbook market. Free, $19 Pro, and $39 Elite tiers with Stripe checkout, magic-link auth via Supabase, tier-gated feature flags, an alert dispatch engine, watchlist presets, and a public API with key management plus usage tracking. The data plane pulls from DataGolf Scratch+ and free Kalshi REST/WebSocket feeds.
AutomateDojo
Done-for-you websites and marketing for martial arts dojos.
A multi-tenant platform that generates branded dojo websites with AI-edited content. Two Claude engines work in concert — Opus for content edits, Haiku for extraction — with an intent classifier dispatching each prompt to the right tool. Per-template fallback registries. Compliance scanning that blocks the model from introducing violations. Version snapshotting for one-click rollback. Plus an integrations layer covering Meta, Stripe, and Twilio.
MarkSearch.ai
Natural-language search across the entire USPTO trademark database.
Plain-English search over the federal trademark register. Maps business descriptions to Nice Classification codes (1–45) automatically. OpenAI embeddings plus Postgres pgvector for semantic similarity. Layered with Meilisearch for full-text speed and Qdrant for vector queries at scale. Includes domain availability checks and lead capture for IP law firms.
SubredMonitor
Real-time Reddit keyword alerts. Multi-user.
Tracks unlimited keywords across any subreddit and fires email plus SMS alerts the moment matches surface. Multi-user auth, persistent SQLite storage with dedupe, configurable polling intervals, a web dashboard with real-time stats. Vercel-deployed Python and Flask — proof that "marketing engineering" doesn't always mean Next.js.
Internal products + investor-facing builds
Builds that aren't paid client work — research, fundraising material, and platforms used as proof-of-concept for new ventures.
GridScout
Datacenter site selection platform.
Interactive map plus scoring engine for evaluating datacenter buildouts. 164,000 sites scored across nine site types — greenfield, industrial, substation-adjacent, mine-reuse, federal, manufacturing, shovel-ready, brownfield, and former military base reuse — using a ten-factor weighted readiness score. Custom weights, location search, hyperscaler tracking. Built originally for an institutional infrastructure investor pitch.
SolarTrack
Commercial and utility-scale solar installation database.
Every commercial (≥25 kW) and utility-scale (≥1 MW) solar installation in the United States, enriched and searchable. Aggregates 26 primary federal and state ingest pipelines, 75 city/county permit portals, and 30 enrichment scripts. Geospatial "near me" search without PostGIS. Built for a solar equipment reseller targeting commercial accounts.
GolfOdds
Live Kalshi-versus-sportsbook arbitrage scanner.
The R&D platform that became SportsBookish. Pulls Kalshi market quotes alongside a 14-book consensus from DataGolf and surfaces pricing edges in real time. Postgres data plane. WebSocket primary with REST fallback. Identified meaningful edges at a major championship — the kind of signal that justified building the productized subscription on top.
Token Opportunity Framework
Capex business plans from the Ribbit Capital token letter.
A static-export Next.js analysis of 150+ companies across the token economy framework. Nineteen detailed capex business plans, each with executive summary, full plan, and financials tabs — TAM, CAGR, LTV-to-CAC ratios, seed asks. Built originally to evaluate which framework opportunities were ready for productization.
Custom HTML pitch decks
Full-screen, navigable, animated.
When a slide deck has to do real work and Google Slides feels apologetic, I build full-screen HTML decks with custom progress bars, prev/next plus keyboard nav, and CSS-animated mockups. Examples: "Distribution Control in the Age of AI" (token framework) and a CalibrateNet pitch for an ADAS calibration network opportunity.
Free marketing tools
Calculators and generators at hyder.me/tools.
Five public tools — Ad Spend ROI calculator, Keyword Match Type generator, UTM Builder, YouTube video search, Backup Power calculator. Each is a single-file HTML page that works offline, no signup required. Direct, useful, and uncluttered — exactly the experience clients ask me to bring to their internal tooling.
What I work in, day to day
Not a brochure list — these are the tools and APIs I have running code against right now, this week.
Languages
TypeScript
Python
JavaScript
SQL
HTML/CSS
Bash
Frameworks
Next.js 16
React
Flask
Tailwind v4
Bootstrap 5
shadcn/ui
Databases & data infra
Supabase Postgres
pgvector
PostGIS
BigQuery
SQLite
Meilisearch
Qdrant
Marketing & ad APIs
Google Ads v23
Meta Marketing API
TikTok Marketing API
Microsoft Ads (SOAP)
GA4
Search Console
Business Profile
AI / ML
Claude (Opus + Haiku)
OpenAI (GPT-4, embeddings)
Intent classification
Vector search
RAG pipelines
Payments, auth & messaging
Stripe v22
Supabase Auth
OAuth 2.0 (6+ providers)
Twilio
Resend
CRM & ops integrations
Airtable
Vonage VBC
Zapier
DataGolf
Kalshi
Infrastructure
Vercel
Cloudflare
GitHub Actions
AWS S3
Cron schedules
Data visualization
Chart.js v4
Recharts
Leaflet
Mapbox
jsPDF
html2canvas
Why custom — not Supermetrics
A six-client agency rolled up into one mega-account needs roughly 25 ad accounts across nine unique data sources. Here's what that looks like in a tools-first world versus a custom one.
| Supermetrics + Looker Studio | Custom build (what I deliver) | |
|---|---|---|
| Annual cost | $7,400 – $21,000 Pro tier with add-ons; higher tier required for BigQuery as destination |
~$0 in tooling Supabase + Vercel free tiers under fair-use limits |
| Annual commitment | Required — no monthly billing | None — you own the code |
| Visualization | Looker Studio templates — generic layout, generic charts | 100% custom per client — branding, drilldowns, exports, your reporting metaphors |
| Refresh cadence | Hourly (Pro tier) or daily (lower tiers) | On-click, every time the user opens a tab |
| Multi-account merges | Rows dumped — you build the merge logic yourself | Built in: 3 Meta accounts unified into one view, 9 Google Ads accounts rolled up with priority brand/non-brand classification |
| Cross-source joins | Custom SQL in BigQuery only (extra cost tier) | Built: GA4 ↔ Google Ads vendor/affiliate attribution, AC ↔ Vonage missed-call attribution |
| Non-API sources | Out of scope (no Vonage call pipeline, no Airtable nuance, no voicemail webhook) | Built (75K call records ingested, $250K+ Airtable rebates synced, voicemail → CRM deal flow) |
| Per-client branding | Limited theme controls | Distinct visual identity per client — gallery gold, financial blue, dark navy, your color |
The catch: Supermetrics still can't deliver about 20–30% of what's running across these dashboards (Vonage pipelines, Airtable workflows, voicemail webhooks, real-time CRM deal creation, custom conversion-action joins, per-client UX). You'd be paying for an incomplete answer.
How working together works
I take on a small number of engagements at any given time — through existing relationships, referrals, and direct outreach from companies I want to work with. Three structures: retainer, project, or advisor. The discovery call is a real conversation about fit, not a sales pitch.