Experiments

Applied AI Experiments

Prototypes and working tools exploring how AI can support different kinds of work.

01 Working prototype

AI Resume Tailoring Workflow

An AI resume workflow that matches prior-resume evidence to a target job, then coordinates drafting, review, and targeted repair.

Problem: Manual tailoring is repetitive, while one-shot AI drafts can miss stronger evidence or overstate experience.
What it does: Uses RAG-informed retrieval to select and rank relevant chunks from prior resumes, then coordinates AI planning, drafting, review, and targeted repair.
Result: Produces an editable resume tailored to the target job description, with weak or unsupported claims surfaced for user review.

02 Public-safe reconstruction

A bounded AI analysis workflow that answers project follow-up questions through controlled tool use and reviewable evidence.

Problem: Reports and presentations cannot easily answer detailed follow-up questions after delivery.
What it does: Lets the model choose among bounded analytical tools to retrieve project evidence and produce a traceable answer.
Result: Makes evidence, methods, assumptions, caveats, and tool activity easier to inspect without replacing the original analysis.