Principal Product Manager
JOHNNY XU
Enterprise Workflows · Gov & Regulated · AI Tooling
Product manager who builds complex workflow and collaboration tools for government and regulated industries — the kind of environments where "move fast and break things" gets you subpoenaed.
I've taken messy, manual processes and turned them into real platforms, launched products from zero, and modernized legacy systems without losing my mind (mostly). I care about shipping things that actually work, measuring whether people use them, and not overcomplicating it.
Recently I've been deep in the AI space — designing responsible AI features for government clients and using Claude Code to prototype faster and help non-engineers help themselves.
this is to show sarah
Committed Traveler
Real trips are built around fixed commitments — a dinner reservation, a temple visit, a hotel check-in. Committed Traveler takes those locked-in events and automatically calculates travel time between them, identifies genuine free windows, and surfaces highly-rated nearby places that actually fit your schedule.
The problem it solves
Most trip planners treat every event equally. Committed Traveler works backwards from what you've already committed to, filling in the gaps intelligently rather than asking you to plan from scratch.
How it works
- Add your start point and end point
- Drop in your commitments with times and durations
- Hit Complete Trip — the app computes travel legs, calculates free time, and fetches ranked nearby suggestions for each gap
Key features
- Color-coded free time blocks — each gap has a matching color across the sidebar, map pins, and search radius circle
- Smart suggestions — scored by rating and review volume, queried across 8 place categories in parallel
- Two ways to add a place — Google Places autocomplete or drop a pin anywhere on the map with reverse geocoding
- Per-block controls — radius slider, category filters (food, drink, culture, nature, shopping, entertainment), and suggestion count
- Bidirectional navigation — click a free time block to pan the map, click a radius circle to scroll the sidebar
- Free time filter — isolate a single leg on the map to reduce clutter
- Printable itinerary — clean export with Playfair Display typography, static route maps per leg, and clickable Google Maps links
- API cost controls — localStorage caching minimizes repeat API calls
- Save / load trips as JSON, with Google Takeout import for personal saved places
Built with
Vanilla JavaScript · Google Maps Platform (Maps JS, Directions, Places, Static Maps, Geocoding) · Vercel · GitHub CI/CD
Home Keeper
A persistent, browser-based tool for tracking everything your house needs to stay maintained. It solves the problem of forgotten maintenance schedules, lost product specs, and not knowing where your shutoff valves are when something breaks.
The problem it solves
Most homeowners keep maintenance schedules in their head, on sticky notes, or nowhere at all. Home Keeper centralizes recurring tasks, product specs, and critical system locations into one persistent interface — so nothing gets missed and everything is findable in an emergency.
How it works
- Add your maintenance tasks with frequencies and last-completed dates
- Link products and consumables to each task with purchase URLs
- Log your house specs and key system locations with photos for quick reference
Key features
- Urgency-sorted task list — color-coded status badges (overdue, due soon, no data) with filterable views
- Linked consumables — products tied directly to tasks with one-click purchase links
- Calendar export — one-click
.icsexport with recurring schedules and day-before reminders, single task or all at once - House data record — structured fields for filter sizes, panel amps, insurance info, plus custom fields for anything else
- Key systems reference — document shutoffs, breakers, and access points with location details and photo uploads
- Pre-seeded data — 12 common maintenance tasks and 8 key system entries as a starting point
- Full persistence — all data saved across sessions with no account or backend required
Built with
React · Persistent browser storage · Dota II-inspired dark UI with angular clip-path cards and Rajdhani typography
Perishables Filter — RimWorld Mod
A community mod for RimWorld that extends the game's stockpile filter system with new categorization options for item management.
The problem it solves
RimWorld's default stockpile filters don't distinguish between items that rot, items that degrade outdoors, or hazardous waste items like wastepacks. Players either over-stockpile into a single zone or spend excessive time manually configuring item-by-item filters. This mod fills that gap with intuitive, broadly applicable filter categories.
How it works
- Enable the mod and open any stockpile zone in-game
- Use the new filter categories to define what the stockpile accepts
- Combine filters to create purpose-built storage zones without touching individual item settings
Key features
- Perishables filter — separates items that require refrigeration from those that don't, enabling dedicated cold storage zones
- Deterioration filter — identifies items that lose condition when left unroofed, helping players protect valuable gear
- Wastepacks filter — isolates hazardous wastepack items into their own category, preventing accidental mixing with general storage
- Mutually exclusive categories — all filter types are designed to be independent and non-overlapping, ensuring predictable behavior
- Lightweight implementation — no new game systems, purely additive to existing UI
Built with
C# · XML · RimWorld Modding API · Visual Studio 2022
- Owned product strategy for core enterprise workflows, turning highly custom customer requests into scalable platform capabilities
- Designed and launched responsible AI features for ~65k government users, navigating strict security and compliance requirements
- Used Claude Code to cut prototyping cycles from a couple times a week to multiple times a day, then ran workshops to get non-engineers using it too
- Introduced Claude Code org-wide — led training that expanded AI adoption across non-engineering teams.
- Led full modernization of multiple legacy platforms — permissions, workflows, data models — without falling into the "rebuild everything exactly the same" trap
- Drove multi-year planning for enterprise deals worth ~$6M ARR across customer, engineering, and services teams
- Regularly played sales engineer, jumping into pre-sales conversations to keep promises grounded in what we're actually building
- Evaluated and selected third-party vendors for document processing, e-signature, and workflow tooling
- Built escalation processes and documentation standards that made stakeholder relationships way less painful
- Hired, onboarded, and mentored across product teams
- Defined product-market fit and built a new enterprise product from concept to launch — true 0-to-1
- Worked directly with early customers to validate workflows, close initial sales, and build credibility with key accounts
- Led early go-to-market efforts including discovery, demos, and implementation planning
- Shipped integrations with GIS and payment platforms
- Moved the product from fully bespoke implementations to a configurable, scalable model that could grow without blowing up delivery risk
- Transitioned off a legacy platform while keeping things stable enough to avoid churn
- Built the company's first product org from scratch — tooling, processes, practices, all of it
- Hired and managed a small product and design team
- Owned products across legislative management, video streaming, and records management
- Led cross-platform integration work after acquisitions, coordinating across video, GIS, payment, and core systems
- Managed scope, timelines, and handoffs across global and offshore teams on complex, multi-team deliverables
- Kept stakeholders aligned and delivery on track in a large, matrixed org where things could easily go sideways
Product strategy & roadmapping, enterprise workflow design, 0-to-1 product development, platform modernization, responsible AI product design, Claude Code & AI-assisted prototyping, stakeholder management (customers, execs, engineering, sales), product analytics & adoption metrics, vendor evaluation & selection, legacy migration planning, cross-team coordination, sales engineering & pre-sales support, hiring & mentoring product teams, integration design (GIS, payments, e-signature, document processing), working in regulated/compliance-heavy environments
- 0→1 Products
- Platform Modernization
- Enterprise Workflows
- Responsible AI
- Product Analytics
- Sales Engineering
- Claude Code
- React / Next.js
- Vue 3
- D3.js
- SQL
- Vercel / Netlify
- Government
- Regulated Enterprise
- Document Management
- Workflow Automation
- e-Signature / GIS
- Northeastern University
- CS / Bioinformatics
- Boston, MA — 2014