Bryan Guo
Summary
Product owner who ships end-to-end — from requirements through production launch — increasingly as a one-person product/dev team applying spec-driven development (requirements → technical design → implementation) in Kiro on AWS (Bedrock, Lambda, API Gateway). CS-trained, with 7+ years turning ambiguous workforce-enablement problems into measurable products across SP-Support, Seller University, and Customer Service & PXT.
Experience
- LearningCat — Own end-to-end (vision, BRD, metadata schema) a service cataloging 50,000+ learning objects across four large LXD teams. Designed the metadata schema — the content → Atlas label → learner-qualification mapping and standardized taxonomies — and aligned it across the four teams, eliminating the free-text naming variants that block automated targeting. This structured data backbone routes the right training to the right associate and is designed as a service ASCEND can pull from to power targeted learning. 3,251 of ~10,000 SP-S active content pieces catalogued to date; drove integration with Nexus/NICE case-routing, with Amazon Learn transcription-service integration on track. Partnered with one data engineer; influences ~80 LXDs and ~35K downstream learners.
- Fan-it — Defined and built Fan-it, a self-serve AI translation product for learning-content localization, as a one-person product/dev team — writing the Lambda logic and standing up the AWS backend (API Gateway, Amazon Bedrock for the AI translation engine, IAM) through spec-driven development in Kiro (requirements → technical design → implementation), steering Claude Opus 4.8 throughout rather than ad-hoc ‘vibe coding.’ Removed ~$100K in annual translation spend and 400+ manual hours and cut turnaround from days to on-demand. Achieved a 3-per-1,000-words issue rate — 90% fewer errors than LELI machine translation and 57% fewer than prior PartyRock tooling. Expanding to new language pairs, video transcription, and additional teams.
- DirecTrain — Launched (CN; scaling to other regions Q3/Q4) a guided publish-first workflow with quality gates that lets ops partners turn a floor-level performance gap into published, assigned training in hours instead of weeks, bypassing the LXD queue. Built solo on the same spec-driven Kiro + AWS stack (Lambda, API Gateway, DynamoDB, S3, IAM).
- AI Enablement — Own a director-level goal targeting a 50% reduction in content-production time; pilots demonstrated 52.4% productivity gains. Facilitated an 8-week AI Bootcamp coaching 16 employees to apply generative AI in daily workflows.
- End-to-end product owner for seller education on the Amazon Seller App (1.3M+ MAU); defined mobile product strategy and roadmap across 10+ engineering and non-technical teams.
- Grew mobile MAU 92.9% YoY, contributing ~$30.6M downstream impact via contextual education; delivered VP/S-team goal contributions at 125% of commitment.
- Transformed a 20-year-old corporate leadership program into a scalable asynchronous solution in Embark New Hire Essentials, used by 100K+ Amazonians (NPS 85, CSAT 4.9, 12K+ reviews).
- Led design of the Business-Tech-Ops executive program, building senior leaders' fluency in ML, distributed systems, and software architecture.
Earlier
Doctoral Researcher, Northwestern University (2012–2019) — Led 3 National Science Foundation (NSF)–funded learning-technology research projects. Drove multi-year, iterative product work on the NetLogo HubNet simulation system, running field studies — observation, focus groups, and interviews with 100+ users — to guide successive improvements. 20+ publications and 5 educational software tools.