Embed in the real workflow
I work close to users and domain experts until the actual constraints, decisions, and measures of success are explicit.
I’m Jerry D. Boonstra, a hands-on AI and founding engineer with a forward-deployed approach: I work from discovery and workflow design through implementation, evaluation, production rollout, and reusable infrastructure.
Latest writingThe Worst Bugs Said Nothing: Building a Tool That Checks AI Plugins →01 / Selected work
An auditable investment-research system connecting policy, evidence, deterministic constraints, agent judgment, and evaluation.
Read case study →Reads a Claude Code plugin before you install it, and publishes the tests it failed.
Read case study →A structured LLM pipeline for transforming career evidence into tailored, ATS-ready resumes.
Read case study →An early machine-operable platform for persistent experiments, lineage, parallel execution, and customer-cloud compute.
Read case study →An ad-supported podcast platform whose audio started at once, because the CDN edge served it in fragments.
Read case study →A career arc from CNET publishing systems through Frequency Networks media infrastructure and a 30-person Jukin technology organization.
Read case study →03 / Career arc
CalmTrader, agentic workflows, and reusable infrastructure for real domain problems.
As CTO, built low-cost, observable LLM application infrastructure.
Led engineering. Built instant-streaming podcast audio at the CDN edge, and the media pipeline behind it.
Modernized the platform and built a 30-person technology organization whose work continued through the company’s later acquisition.
Led DevOps and API work for a video channel-creation and global content-distribution platform.
Built publishing systems, then led site and platform engineering for major consumer properties.
04 / Selected writing
05 / Contact
I’m open to Forward Deployed Engineer, founding-engineer, and staff/principal IC conversations with teams building technically ambitious AI products.