Skip to content
Woody Kim

Woody Kim

Design Engineer

I'm a design engineer and founder of Midnight Health. I build secure cross-platform tooling for health-conscious users.

Previously Design and Strategy at Kessel Run, DIUx's first success story, which modernized U.S. digital infrastructure.

Designed at a.team, Bubbles, and ActivisionBlizzard.

Based in New York City.

Reach out

Selected Work

Three stories about designing the system around the screen: product decisions, implementation constraints, and what shipped.

iOSTodayCheck-in
Mobile webTodayHistory
TabletPlanSignals
DesktopPlanTrends
One workflow, four contextsNavigation and density adapt to the device; the product model stays recognizable.
TodayOne useful planDo less, understand why.
Review what changed
The day resolves to one planExisting signals are reduced to a short plan with supporting context available on demand.
SourcesApple HealthDaily check-in
Explicit consentSource and scope stay visible
Limited outputDaily planPersonal history
Trust is part of the workflowConsent, source, and scope are shown where the user makes a decision.

1 / 3

In production2026–Present

Midnight Health — One Product System Across Contexts

A secure daily health workflow designed as one system across iOS, desktop, tablet, and mobile web.

Role
Founder, product design engineer

Disclosure
Public product. The diagrams omit personal health data and account details.

Read the case
Source context
AI draft
Human review
Accepted output

Nothing advances without an accountable decision.

Human review is the workflowSource context, draft comparison, and explicit acceptance keep ownership with the operator.
EvaluationPrompt + contextObserved outputFailure modeRegression check
Feedback becomes evaluation evidenceTeams can connect a poor result to the prompt, context, and failure mode that produced it.
Shared coreContext · draft · review · feedback
PlanningResearchOperations
A shared core, adapted by teamCommon review states support different writing jobs without fragmenting the product model.

1 / 3

Sanitized2024–2025

AI Writing Tools at Scale

Human-in-the-loop writing workflows designed for accountable use across enterprise product teams.

Role
Product designer, UX strategist

Disclosure
Sanitized case study. No internal interfaces, operational data, or restricted workflows are shown.

Read the case
Foundations language · type · statesComponents inputs · evidence · reviewProduct workflows real decisions, ready to ship
A system grown from product needsFoundations, components, and workflows were defined together instead of as separate deliverables.
Before
Complete task
Leave product
Find requirement
Return
After
Complete task
Review evidence
Confirm
Compliance in the working pathRequired evidence and review appear at the decision point without turning the whole product into a form.
One product truthWorking product state
Users understand the next actionEngineering understands the stateFounders can demonstrate the value
One product truth, two audiencesThe interface serves the user while giving founders a credible, demonstrable product narrative.

1 / 3

Sanitized2022

Series A Product & Compliance UX

Product systems that helped early-stage teams make complex, compliance-adjacent software usable and credible.

Role
Product designer

Disclosure
Anonymized multi-client case study. Company-specific screens, names, and fundraising details are omitted.

Read the case

Experience

Education

  • MIT CSAIL

    Deep Learning S.191 (Sabbatical)

    2026
  • MIT Phantom AI Accelerator

    USAF x MIT

    2023
  • University of Southern California

    B.A. Design + Entrepreneurship

    2016–2020

Skills

  • Product UI
  • Design systems
  • Working prototypes
  • Cross-platform craft
  • AI interfaces

Tools

  • Claude Code

    Main coding

  • Magic Patterns

    Design system

  • Figma

    Interface decisions

  • Next.js

    Web apps

  • Expo · React Native

    Native iOS

  • TypeScript

    Shared contract packages

  • Supabase

    RLS

  • promptfoo

    Evaluation gate for deterministic workflows

  • Antigravity

    Evals