Product leader · Tech entrepreneur
I’m John Wen. I build at the intersection of data, AI, agents, and developer experience—leading enterprise products and turning my own ideas into growing businesses.
01 / Approach
“The best product work makes a difficult future feel obvious, useful, and close.”
02 / Selected work
Four IBM product chapters: from applied AI and cloud infrastructure to the developer and agentic interfaces behind modern data systems.
Agentic data integration
Took an agentic data product from concept to stable 1.0.0 in six months, giving customers a natural-language interface to build and operate pipelines. The launch opened enterprise PoCs and now supports customers in production.
Pipelines as code
Built the code-first foundation for streaming, batch, and unstructured pipelines from zero to production across 8+ releases—giving teams automation at scale and downstream agents a reliable control plane.
Cloud platforms
Owned go-to-market, pricing, and packaging for managed Kubernetes and container registry services—growing an anchor-client program from $36M to $48M ARR and increasing registry revenue 5×.
Applied enterprise AI
Led the AI roadmap for IBM’s governance, risk, and compliance platform—driving 300% feature adoption, reducing analyst revision time by 20%, and generating $5M+ in license and services revenue.
03 / Ventures & projects
I like the whole loop: finding the wedge, building the product, earning attention, reading the data, and making the next call.
Cofounder & CEO
A pay-by-the-minute Minecraft server hosting company built around a simpler promise: play when you want, pay only when the server is running.
Founder & Builder
An AI-powered deal intelligence platform that turns private-equity data rooms into cited, decision-ready insights across active deals.
Georgia Tech · M.S. Analytics
Applied work across product analytics, natural-language processing, statistical learning, and responsible model evaluation. GitHub repositories and deeper write-ups are coming next.
Built an end-to-end analytics pipeline that turns raw Instagram comments into per-creator briefs with audience themes, content opportunities, and clear growth or monetization actions.
Tested whether an AI agent’s failure risk could be predicted before execution from task and model metadata—while treating unusually strong results as a data-quality question, not an automatic victory.
The project stories are live now. Code and deeper methodology will be linked when the repositories are ready.
04 / Agent system
Private research product · Designed, built, and deployed
Hermes Stock Agent · “Ryan”
I built a private stock-research agent that turns portfolio context and live market evidence into concise, owner-first decision support. The system distinguishes facts from model judgment, preserves uncertainty, and never places trades.
Dynamic portfolio context, market data, company fundamentals, filings, news, and source health.
Bounded evidence packets with freshness checks, contrary evidence, and explicit decision gaps.
Strict decision schemas and fail-closed publishing prevent incomplete research from becoming a new call.
The same accepted state powers concise Discord briefs, alerts, history, and a private dashboard.
One 1–3+ year ownership lens separates business-thesis health from short-term timing and technical risk.
Transition-based monitors surface meaningful position, filing, news, and trend changes without repeating unchanged alerts.
A mobile-first React and Express application prioritizes owner alerts, market discovery, the best three actions, positions, and evidence.
Structured validation, source-health checks, deterministic rendering, Docker deployment, and rollback-safe operations keep the system trustworthy.
05 / About
I help teams find the useful product hiding inside a complex technical system. That means listening closely, getting fluent in the details, and making a clear bet on what should exist next.
That instinct continues after work. I cofounded Campfire Hosting to learn the operating loop firsthand, and built Vyntic to explore how cited AI can reshape private-equity diligence.
A conversation about analytics, entrepreneurship, teaching, and the curiosity that shaped my path from Babson to product.
Read the profile2024 — Now
Campfire Hosting
Growing a profitable pay-by-the-minute game-server business through product, brand, AI-powered marketing, and usage analytics.
2025 — Now
IBM · watsonx.data integration
Building agentic data integration products, including the Unified Python SDK and MCP Server.
2022 — 2025
IBM Cloud · Containers
Led product and go-to-market work across Kubernetes, OpenShift, and container registry services.
2020 — 2022
IBM Data & AI · OpenPages
Set the AI product roadmap and brought machine-learning experiences into enterprise risk workflows.
06 / Thinking
2026
The next generation of agents will assemble live, task-specific data pipelines on demand—and retire them when the question is answered.
2026
When intent connects directly to execution, teams can build, edit, and migrate data workflows without the usual translation layer.
2025
Code gives people, LLMs, and agents the same composable interface for creating and maintaining modern data systems.
2022
An early look at how predictive models can improve data quality, guide risk analysts, and bring AI directly into enterprise GRC workflows.
07 / Beyond the roadmap
Outside the day-to-day, I’m usually exploring the systems behind better decisions—and building small things to test what I learn.
Agents, automation, and the interfaces that make them genuinely useful.
Studying durable businesses, long-term ownership, and decision quality.
Graduate work in machine learning, modeling, and turning messy questions into evidence.
Selected moments
08 / Connect
Let’s compare notes.