Yiming X.

I help startups and businesses turn ideas into working products quickly, combining product strategy, AI-assisted development, automation, deployment, customer discovery and rapid iteration. Before this, I spent six years in enterprise software delivery in regulated industries. I started at Accenture as a technical business analyst, building SQL data models and Tableau dashboards for pharmaceutical manufacturing. At Pfizer I became a Solution Owner, leading teams of 10+ engineers and designers through discovery, requirements, and production deployment in a GxP environment. I also co-invented a patent for predictive workflow software. I am not a traditional software engineer. I am an AI-native product builder and operator who takes a business problem from concept to a deployed, continuously improving product, and then goes and finds the customers for it. I work daily in Claude Code and OpenAI Codex as a practical development workflow: prototype, build, test, deploy, troubleshoot, iterate. That includes building the skills, MCP servers and agent workflows those tools run on, not just prompting them. Products I have built and operate: Hospital Optics — a staffing contract intelligence platform for hospitals. The source agreements weren't in any database, so I built an agent pipeline that ran the FOIA campaign itself: finding records contacts, filing requests over Gmail, following up, answering agency questions, and processing what came back. 400+ requests across 30+ states, largely hands-off. I then built the pipeline that OCRs the agreements, extracts every clause with a page citation, and verifies each finding against the source page. On top of that, an MCP interface where users ask open-ended questions about their contracts or how they compare to peers, and get answers back in whatever format they need. Acadexis — a higher-education compensation intelligence platform built on large-scale employee and salary datasets: data pipelines, normalization, benchmarking, AI-assisted workflows, and customer-facing applications that help institutions make better compensation decisions. For both, I built and run the go-to-market on a fully custom outreach and CRM system: email drafting through the Gmail API with pre-send verification checks, an engagement funnel and contact records in Postgres, and LinkedIn outreach tracked alongside it. Capabilities: - Rapid MVP and prototype development - LLM-powered applications, AI agents and agentic workflows - Custom MCP servers and tools, Claude skill and project architecture - Document processing, structured extraction and citation-verified outputs - Public records acquisition at scale (FOIA strategy, request tracking, follow-up) - Data pipelines, PostgreSQL, S3-compatible storage, analytics dashboards - Python, FastAPI, Next.js, React, REST APIs - Deployment on Vercel and Fly, backend services and database infrastructure - Workflow and business-process automation - Customer discovery, outbound email and LinkedIn outreach, custom CRM and outreach tooling - Product strategy and technical product management My strongest skill is end-to-end ownership. Instead of separate people to define requirements, build, deploy, talk to customers and decide what comes next, I run that whole loop myself. I am most effective where the problem is still ambiguous and the job is to work out what should be built, build it fast, put it in front of users, and iterate on what comes back. If you have an idea, an inefficient workflow, an early-stage product, or a problem that software, automation or AI could solve, I can take it from concept to prototype to deployed product to customer feedback to continuous improvement.