Our QA teams rely on paid validators plus dev tools to review third-party tags. That works, but it’s license-bound and slow when you just want to sanity-check a tagsheet in one place.

• Problem: Core checks (HTTPS, dimensions, weight, click URL, landing page status) lived across multiple tools and a vendor platform. There was no lightweight way to scan a full tagsheet quickly across vendors.
• Solution: Built a browser-based page where ad ops can paste a single tag or upload a tagsheet, auto-detect tags from major vendors, render a live preview, and surface key metrics like load time, requests, size, dimensions, vendor, placement ID, and landing-page response.
• Impact: Shipped a working proof-of-concept that recreates our most-used AdValidation checks in-house. While session-only, TagSafe shows a clear path to reducing validator licensing costs with a focused internal tool.

Some custom video placements on our site can’t accept VAST tags—we need a standalone MP4 and 1×1 trackers. When campaigns were sold months in advance and only a VAST tag was delivered, we had no way to recover the assets.

• Problem: Agencies often send only a VAST tag, leaving no time to request new files. Extracting media and trackers manually from raw XML or network tools is slow and brittle.
• Solution: Built a browser tool where anyone can paste a VAST tag, preview its metadata, download the MP4 asset, and export impression/click trackers as 1×1 files. Works as both an extractor and a lightweight validator.
• Impact: Used by ad ops and account teams to keep campaigns live without going back to clients. Allowed us to run Meta and Amazon units on formats that normally can’t accept VAST.

Brand-safety reports from IAS arrive as long lists of article URLs and hits, but our ad server targets on internal page IDs. Manually mapping each URL to its pageID in source code doesn’t scale.

• Problem: Excluding “unsafe” articles meant opening every URL, finding the pageID in the markup, and copy/pasting it into a blocklist. Hundreds of URLs could turn into hours of repetitive work and easy mistakes.
• Solution: Built a tool that ingests a spreadsheet of URLs, normalizes and dedupes them (including UTM variants, games pages, and other non-ID paths), rolls up impression counts, lets users set a minimum hits threshold, then calls a scraping worker to return pageIDs for only the high-impact pages.
• Impact: Turns a multi-hour CMS task into a paste-and-download workflow used regularly by ad ops. Typical IAS exports shrink from tens of thousands of URLs to a few dozen pageIDs that cover the vast majority of unsafe impressions for a campaign.

Helps keep scope, priority, and progress legible as projects evolve.

• Problem: My side projects tended to sprawl across notes, boards, and docs. I needed a tool help me stay organized & on track as my tools grew in scope, while being able to present my project, all in one place.
• Solution: Built a browser-based planning tool that combines PRD authoring, feature roadmapping, and phase-level progress into one visual system. Workbench supports fluid drag-and-drop, inline editing, undo/redo, and Markdown export—making it easy to plan, iterate, and explain a project without leaving the page.
• Impact: Replaced scattered notes and task boards with a single source of truth for personal builds. Helped maintain clarity and momentum on long-running side projects, and made it easy to present a high-level product story to others without slides or external docs.

Our QA teams rely on paid validators plus dev tools to review third-party tags. That works, but it’s license-bound and slow when you just want to sanity-check a tagsheet in one place.

• Problem: Core checks (HTTPS, dimensions, weight, click URL, landing page status) lived across multiple tools and a vendor platform. There was no lightweight way to scan a full tagsheet quickly across vendors.
• Solution: Built a browser-based page where ad ops can paste a single tag or upload a tagsheet, auto-detect tags from major vendors, render a live preview, and surface key metrics like load time, requests, size, dimensions, vendor, placement ID, and landing-page response.
• Impact: Shipped a working proof-of-concept that recreates our most-used AdValidation checks in-house. While session-only, TagSafe shows a clear path to reducing validator licensing costs with a focused internal tool.

Many client assets arrive with white edges. Policy requires a visible border so ads aren’t confused with article imagery, but updating each size manually slows launches.

• Problem: Adding borders meant re-requesting assets or opening each creative in Photoshop/Canva. For several concepts across 6+ sizes, that easily became a 10–20 minute chore per campaign.
• Solution: Built a browser tool where anyone can upload standard display assets, apply a policy-compliant border in one click, and download renamed files that clearly indicate the border has been added.
• Impact: Turned a license-dependent, ~1-minute-per-asset step into a ~15-second process for the full creative set. A typical once-a-week border job now takes under a minute instead of up to 20.

Many local advertisers bring photos and copy but don’t have design resources. Simple display campaigns can stall while someone with Photoshop/Canva builds out every size by hand.

• Problem: Producing 970x250, 300x600, 320x50, 300x250, and 728x90 units from the same concept meant recreating layouts in a design tool, one size at a time. That required licenses, design skills, and lots of copy-paste work.
• Solution: Built a browser layout engine where users paste in headline, description, CTA, and upload images once. AdBuilder auto-renders the full set of standard sizes, lets you tweak copy per size with live previews, and exports final creatives as PNGs.
• Impact: Demonstrates how non-designers could generate full creative sets without art software, reducing friction for simple campaigns and setting the stage for future HTML5, responsive, or video-based exports.

Some custom video placements on our site can’t accept VAST tags—we need a standalone MP4 and 1×1 trackers. When campaigns were sold months in advance and only a VAST tag was delivered, we had no way to recover the assets.

• Problem: Agencies often send only a VAST tag, leaving no time to request new files. Extracting media and trackers manually from raw XML or network tools is slow and brittle.
• Solution: Built a browser tool where anyone can paste a VAST tag, preview its metadata, download the MP4 asset, and export impression/click trackers as 1×1 files. Works as both an extractor and a lightweight validator.
• Impact: Used by ad ops and account teams to keep campaigns live without going back to clients. Allowed us to run Meta and Amazon units on formats that normally can’t accept VAST.

Brand-safety reports from IAS arrive as long lists of article URLs and hits, but our ad server targets on internal page IDs. Manually mapping each URL to its pageID in source code doesn’t scale.

• Problem: Excluding “unsafe” articles meant opening every URL, finding the pageID in the markup, and copy/pasting it into a blocklist. Hundreds of URLs could turn into hours of repetitive work and easy mistakes.
• Solution: Built a tool that ingests a spreadsheet of URLs, normalizes and dedupes them (including UTM variants, games pages, and other non-ID paths), rolls up impression counts, lets users set a minimum hits threshold, then calls a scraping worker to return pageIDs for only the high-impact pages.
• Impact: Turns a multi-hour CMS task into a paste-and-download workflow used regularly by ad ops. Typical IAS exports shrink from tens of thousands of URLs to a few dozen pageIDs that cover the vast majority of unsafe impressions for a campaign.

IAS brand-safety reports arrive as hundreds of URLs and impression counts, but GAM blocklists rely on keywords and key-values—not URLs. Translating one into the other was slow, inconsistent, and highly manual.

• Problem: Ad ops had to review each URL, decide what it meant, extract terms by hand, and build a negative keyword list in spreadsheets—an error-prone task that varied greatly by who did it.
• Solution: Built a tool that ingests URLs, extracts terms between hyphens, dedupes them, surfaces their frequency and impression impact, and lets users build a custom blocklist interactively—adding and removing terms with one click.
• Impact: Converts a subjective spreadsheet process into a consistent, 5–10 minute workflow. Blocklists can now be built quickly from real impact data and exported directly for GAM exclusion targeting.

Brand-safety keyword lists often arrive as long columns from Excel exports. Cleaning them manually— trimming punctuation, deleting duplicates, and avoiding stopwords—takes time and easily introduces errors.

• Problem: Turning thousands of spreadsheet rows into a usable negative keyword list meant hunting duplicates and checking for stray formatting issues by hand—sometimes across multiple columns.
• Solution: Collate auto-detects keyword columns, removes duplicates, strips punctuation, flags high-risk stopwords, sorts terms alphabetically, and outputs a clean comma-separated string or .txt file.
• Impact: Turns a 10–15 minute Excel chore into a five-second drop-and-copy step. Used frequently to prepare brand-safety keyword lists without needing Excel or manual clean-up.

Helps keep scope, priority, and progress legible as projects evolve.

• Problem: My side projects tended to sprawl across notes, boards, and docs. I needed a tool help me stay organized & on track as my tools grew in scope, while being able to present my project, all in one place.
• Solution: Built a browser-based planning tool that combines PRD authoring, feature roadmapping, and phase-level progress into one visual system. Workbench supports fluid drag-and-drop, inline editing, undo/redo, and Markdown export—making it easy to plan, iterate, and explain a project without leaving the page.
• Impact: Replaced scattered notes and task boards with a single source of truth for personal builds. Helped maintain clarity and momentum on long-running side projects, and made it easy to present a high-level product story to others without slides or external docs.