# Nelson Spencer Portfolio I build AI-native products, thoughtful interfaces, and internal tools that help teams work better. ## Focus Areas - AI-NATIVE PRODUCT ENGINEERING FOR NEW PRODUCTS, AGENTIC TOOLS, AND SOFTWARE WHERE AI IS THE EXPERIENCE. - DESIGN ENGINEERING FOR AMBIGUOUS FLOWS: PROTOTYPES, INTERFACES, AND POLISH THAT MAKE NEW BEHAVIOR FEEL USABLE. - AI ENABLEMENT, AUTOMATION, AND GTM ENGINEERING FOR INTERNAL TEAMS WHERE THE PROBLEM IS THE POINT. ## Work ### 01. Recruiting Operations Platform Tags: Client Project Outcome: 20+ hrs/week reclaimed AI enablement for a GTM-facing recruiting workflow, turning manual role sourcing and publishing into a daily operating system. Problem: A team was manually collecting, deduplicating, reviewing, and publishing jobs across a fragmented weekly workflow. Built: - Designed the review experience and built the ingestion, deduplication, publishing, performance tracking, and newsletter workflows behind it. Result: Publishing moved from weekly to daily, the team reclaimed 20+ hours per week, and the workflow became a single operational system. Architecture: TypeScript full-stack platform on Node.js with Convex, Upstash Redis, Trigger.dev background jobs, Firecrawl/OpenRouter ingestion, and Vercel deployment. Links: - Live site: https://jobs.theplusonecollective.com/ ### 02. Community Hiring Platform Tags: Client Project Outcome: ~300 monthly active users Product engineering for a Slack-native hiring workflow, moving role management, applicant tracking, and follow-up out of threads into a shared tracking system. Problem: A community was managing hiring through Slack threads, which made applicant tracking, follow-up, and accountability hard to maintain. Built: - Built the product surface and workflow layer: role management, applicant tracking, Slack-native posting, member authentication, and automated notifications. Result: The platform supports roughly 300 monthly active users and gives the community a structured system of record around hiring activity. Architecture: Next.js hiring platform with Supabase PostgreSQL, Drizzle ORM, Slack API integrations, and Clerk authentication. ### 03. Website Feedback Intake Tool Tags: Internal Tool Outcome: $100+/mo in tooling replaced Design engineering for an internal product workflow: screenshot capture, annotation, and Linear routing in one lightweight tool. Problem: As I was building products for clients, I needed an easy and efficient way for them to provide feedback. The existing tools felt heavier than the problem, so I wanted to see if I could build a simpler version myself. Built: - Designed and built a Chrome extension that captures screenshots, supports annotations and redactions, collects bug and enhancement requests, and routes feedback directly into Linear. Result: Replaced the third-party tool in my workflow and kept client feedback flowing directly into the development process. Architecture: Chrome MV3 extension with a Convex backend, screenshot storage, domain-based Linear routing, and email notifications. Links: - Chrome Store: https://chromewebstore.google.com/detail/augi-feedback-tool/eecckpifojdkdoodnmiiikkokipcffcb ### 04. Metro-North MCP Server Tags: Open Source Project Outcome: Public on npm ยท v2.1.0 AI-native product engineering for agents: an open-source MCP server that gives personal AI tools real Metro-North transit context. Problem: I wanted my personal AI agent (Hermes) to understand my Metro-North commute using real schedule, station, alert, and departure data. Built: - Designed typed tool contracts and built an MCP server using public Metro-North feeds for schedules, service alerts, realtime departures, station lookup, and trip planning. Result: The project is now a public npm package with tests, CI, documentation, and support for MCP-compatible clients. Architecture: TypeScript MCP server on Node.js with SQLite + Redis caching, Dockerized, powered by MTA GTFS and real-time feeds. Links: - GitHub: https://github.com/NelsonSpencer/metronorth-mcp - npm: https://www.npmjs.com/package/metronorth-mcp - Release v2.1.0: https://github.com/NelsonSpencer/metronorth-mcp/releases/tag/v2.1.0 ### 05. WhatsApp Reporting Prototype Tags: Consulting Project Outcome: Structured capture in-channel Design engineering prototype for field operations, turning unstructured WhatsApp partner communication into structured workflows. Problem: A social enterprise coordinated field partner updates and support through WhatsApp, but important operational data was trapped in unstructured chats. Built: - Turned stakeholder research and workflow mapping into WhatsApp Flows that captured structured reports and support requests inside the channel partners already used. Result: The prototype showed how unstructured partner communication could become structured operational data without forcing a new tool into the workflow. Architecture: WhatsApp Flows prototype for structured reporting, partner support workflows, and operational data collection, informed by human-centered research. ### 06. AI Enablement Training Tags: Workshop Outcome: Working automation in 3 days AI enablement training that helped an operations team build fluency and design practical AI-powered workflows. Problem: A small ops team wanted to build AI fluency and practical automations to multiply their capacity. Built: - Designed and led a three-day workshop where participants mapped their own use cases, then built a customer-support workflow that triaged emails and routed work into Asana. Result: The team left with a working automation, a reusable workflow pattern, and more confidence applying AI to day-to-day operations. ## References ### The Plus Side Team Context: Recruiting Operations Platform > When we first started The Plus Side Job Board, the entire process was manual. Every week we spent hours combing through different job boards, selecting roles, and copying and pasting them into our system. It was tedious and time-consuming. > > Meeting Nelson was a significant turning point. He quickly understood our workflow and built an automated solution that streamlined the entire process. What stood out most was how easy he was to work with: he took a complex situation and simply handled it. There was very little back-and-forth, the rollout was seamless, and the system was up and running faster than we expected. > > And the icing on the cake: it looks fantastic. The final product is not only efficient, but beautifully designed and easy to use. > > I would highly recommend Nelson to any organization or business trying to figure out how AI or automation can improve their processes, systems, or operational workflows. ### Alvin Belt Context: AI Enablement Workshop > The way everything was broken down made it very easy to follow and navigate the terms. Nelson helped me change my way of thinking about the different ways to leverage automation tools and AI to work more efficiently. I really enjoyed working through things while being guided. It made me even more eager to do it independently. VP of Operations, AC Brands ### Nicole Iturbe Context: AI Enablement Workshop > I went from knowing very little about how AI tools could be leveraged in my day to day to actively thinking about different automations that could be used to save us time as a part of a small team with big goals. Senior Director of Consumer Strategy and Operations, AC Brands ## Links - Portfolio: https://portfolio.ns.life - Ways to work together: https://ns.life/work - LinkedIn: https://www.linkedin.com/in/nelsonspencer - GitHub: https://github.com/nelsonspencer - X: https://x.com/n_spencer - Email: mailto:ns@ns.life - ns.life: https://www.ns.life/