# Paul Yu — Senior Analytics Engineer > I turn messy source data into reliable models, reporting, and AI tools. I also work with the humans who use them, because good engineering only matters when it helps someone make a better call. - Location: Toronto, Canada - Human website: [https://paul-yu.com](https://paul-yu.com) - LinkedIn: [https://www.linkedin.com/in/paulyu01](https://www.linkedin.com/in/paulyu01) - GitHub: [https://github.com/pyxn](https://github.com/pyxn) - Structured portfolio: [https://paul-yu.com/agent/portfolio.json](https://paul-yu.com/agent/portfolio.json) ## Selected work 1. [33.3× Faster Dashboards](https://paul-yu.com/agent/case-studies/dashboard-performance-engineering.md) — How one Sigma dashboard went from ten widget-level query paths to a shared data model — and how to make the same kind of change without changing what the numbers mean. 2. [Trusted AI Answers For Leadership](https://paul-yu.com/agent/case-studies/semantic-layer-for-ai.md) — How a governed data layer taught Cortex the grain, joins, and metrics — and stopped a fluent guessing machine from reaching the humans who needed it. 3. [16M Rows, Precisely Orchestrated](https://paul-yu.com/agent/case-studies/warehouse-pipeline-orchestration.md) — How one scheduled root and a Snowflake task graph replaced guessed wait times — and kept incomplete refreshes from reaching dashboards. 4. [700k-Scale Identity Resolution](https://paul-yu.com/agent/case-studies/creator-identity-resolution.md) — How graph-based identity resolution replaced a manual linking queue that could not keep up with tens of thousands of new creators each month. 5. [One Secure Data Policy](https://paul-yu.com/agent/case-studies/sigma-row-level-security.md) — How a four-switch Sigma policy replaced hard-coded workbook access across hundreds of clients with composable organization, agency, division, and brand permissions. 6. [99% Fewer Data Errors](https://paul-yu.com/agent/case-studies/campaign-data-quality.md) — How a checker inside a shared spreadsheet stopped bad campaign links at the moment of entry — and broke an endless correction loop before reporting. 7. [Answers for 30+ Teammates](https://paul-yu.com/agent/case-studies/data-knowledge-and-lineage.md) — How a versioned GitHub atlas replaced word-of-mouth warehouse knowledge — so 30+ teammates and their agents could stop chasing the person who knew. 8. [Criticism Into Creative Direction](https://paul-yu.com/agent/case-studies/ai-audience-insight-loop.md) — How sentiment, LLM topic analysis, and human review turned scattered criticism into a simple “do this, not that” brief, then checked the lesson against a later campaign with 86.8% positive sentiment. 9. [75% Less Storage. 84.6% Less Peak Memory.](https://paul-yu.com/agent/case-studies/rolefarer-storage-efficiency.md) — How content hashes cut storage for repeated job text by 75% while a bounded publisher cut peak VPS memory from about 7 GB to 1.078 GB, without losing job history. 10. [146 Agent Tools. One Balanced Ledger.](https://paul-yu.com/agent/case-studies/clovis-ai-action.md) — How AI handles changing statement formats, match rules classify the known cases, and a double-entry ledger turns scattered financial data into a personal CFO. ## Career ### Analytics Engineer, Business Intelligence — Viral Nation Mar 2025 to present I work as an analytics engineer and solution architect for reporting and AI. I build the shared Snowflake data layer behind dashboards and AI products. A 16-million-row model made median Sigma queries 33.3× faster, from 9.6 seconds to 0.29 seconds, with zero system errors across 4,116 launch queries. I also built data orchestration across four creator programs that cut month-long approval work to 2–3 days, a governed AI layer with 24 reusable metrics and 32 verified questions, and identity matching that reduced fragmented creator records by 36.6%. ### Supervisor, Marketing Science — Touché! Aug 2023 to Mar 2025 I was a measurement leader for major national client accounts. I led measurement for Canadian Tire, WestJet, HSBC, and Canada Post, connecting KPIs, data, and recommendations to media decisions. I built a Domo, Python, and SQL pipeline that automated competitive intelligence reporting, cut production from 1–2 months to one week, and generated more than $200K in new billings. I also managed three analysts, coached the team in SQL, Python, dashboards, and privacy-safe data clean rooms, and won Creative of the Year for marketing science automation. ### Senior Analyst, Marketing Science — Touché! Oct 2021 to Aug 2023 I was a hands-on analytics builder who turned repeat measurement work into reusable systems. I built Observer, a Domo data-governance system that automated lineage, ownership, and quality alerts. It cut anomaly detection from days to minutes and saved 20 hours of manual monitoring each week. I also built reusable SQL, Python, and R analysis patterns for customer journeys, conversion, loyalty, audience overlap, and category movement, giving the team a tested starting point for repeat client questions. ### Data Analyst, Data Science — Starcom Mar 2019 to Nov 2019 I was a data analyst focused on making campaign and billing data trustworthy. I combined campaign and finance data, checked invoices, and followed up with media partners when something did not match. After spotting a slow manual process, I learned VBA and built an invoice-matching tool in two weeks that saved the billing team more than 20 hours each month. ## Reviews > Paul was one of the most dedicated and reliable teammates I've ever worked with. He has a gift for explaining complex topics in a clear and didactic way, making collaboration seamless and productive. > > Former teammate · Touché! · LinkedIn recommendation, 2025 > Working with Paul was like watching a masterclass in marketing science automation. He doesn’t just believe in clean, precise data—he builds the systems that make it happen, over and over again. > > Former teammate · Touché! · LinkedIn recommendation, 2025 > His work has not only improved data accuracy by nearly 99 percent but has also shifted the internal culture toward one of high trust and data-backed confidence. > > Manager · Viral Nation · Year-end review, 2025 > Paul’s ability to solve ambiguous technical challenges while maintaining a selfless, team-first mindset makes him an invaluable asset to the organization. > > Manager · Viral Nation · Year-end review, 2025 ## Industry experience - **AI and software:** Product reporting, creator programs, competitive analysis, and AI-assisted insight. Examples: OpenAI, Meta AI, Elastic, Manychat - **Social platforms:** Campaign measurement, content analysis, reporting systems, and data quality. Examples: Meta, Facebook, Instagram - **Consumer technology:** Launch reporting, audience analysis, social listening, and QBR measurement. Examples: Meta Quest, Ray-Ban Meta, Roborock - **Household and personal care:** Multi-brand reporting, benchmarks, paid media analysis, and quality checks. Examples: Kimberly-Clark, Kleenex, Huggies, Cottonelle, Pull-Ups, Goodnites, Poise, Depend - **Retail and ecommerce:** Customer journeys, conversion, loyalty, campaign reporting, and sales analysis. Examples: Canadian Tire, Vans, Polywood, Oakley - **Automotive:** Dealer performance, target automation, campaign measurement, and reporting design. Examples: Audi, Volkswagen, Stellantis, Chrysler, Dodge, Ram - **Financial services:** Privacy-safe media exposure, reach, frequency, sequence, and conversion analysis. Examples: HSBC - **Travel and logistics:** Digital behavior analysis, reporting, and measurement infrastructure. Examples: WestJet, Canada Post - **Food and beverage:** Portfolio measurement, campaign analytics, and consumer behavior research. Examples: Diageo, Campbell's - **Education:** Digital analytics implementation and measurement planning. Examples: McMaster University ## Featured clients OpenAI, Meta, Facebook, Instagram, HSBC, Canadian Tire, WestJet, Canada Post, Stellantis, Volkswagen