Documentation
This is the workflow behind im-khang.com: a repeatable, recruiter-visible process that turns raw case-study work into trusted, documented, deployed analytics products.
Modern analytics teams increasingly use AI agents to compress work across requirements, code, testing, documentation, and deployment. The differentiator is not “using AI”; it is designing controls so agent speed still produces auditability and sound business judgment.
I use Hermes Agent as a task-execution layer while keeping ownership over metric definitions, business interpretation, caveat boundaries, and final recommendations.
Same structure expected from strong data teams: discovery, analysis, validation, narrative, delivery, review, iteration.
Hermes helps transform broad goals into decision questions, stakeholder assumptions, and acceptance criteria.
Datasets are profiled for fields, grain, missingness, leakage risk, and unavailable operational variables.
SQL/Python notebooks and scripts generate metrics, rankings, cohorts, and reproducible aggregate artifacts.
Hermes accelerates test writing, static checks, link checks, and local smoke tests before any publish.
Human review separates proxy from direct evidence, association from causality, and insight from policy action.
Open Design creates recruiter-readable pages from accepted analysis artifacts and narrative constraints.
Hermes/gh script creates reviewable PRs; Vercel creates previews; main branch publishes to im-khang.com.