SYS_ACTIVE // KHUSHI SHAH

I build products that make complicated systems easier to use, trust, and act on.

From AI intelligence tools to security workflows and operational systems, I like finding the real bottleneck, deciding what matters first, and building the version that makes someone's work meaningfully easier.

EXPLORE PROJECTSSEE HOW I THINK

Featured work

2 LIVE · 2 IN PROGRESS

Each case study is a decision map: how a complex system was made clearer to use, trust, and act on.

OPERATIONAL SYSTEMS

Spotlight Security

From scattered security signals to a usable path to action.

Security Product Engineering — brought together host, firewall, and customer context into prioritized findings with an actionable remediation path.

DISPARATE TOOL SIGNALS→CLEAR PRIORITY→ACTIONABLE REMEDIATION
READ CASE STUDY →
HOST AGENTSFIREWALL CONFIGNETWORK CONTEXTNORMALIZE + DETECTPRIORITIZED FINDINGACTION + FOLLOW-THROUGH

AI WORKFLOW DESIGN

RiskLink

Designing and Engineering a voice-led cyber-risk assessment workflow.

AI Product Engineering — shaped a voice-based intake that turns conversation into structured risk reports for cyber insurance providers

NATURAL LANGUAGE INPUT→RISK MAPPING→HUMAN HANDOFF
READ CASE STUDY →
"Tell me about your...""Can you clarify...""One more thing..."JSON OUTPUTGENERATED REPORTTEAM FOLLOW-UP →

Up next

Case studies publishing as they go live.

PRODUCT LOGIC + TRUST

Vantage / LETSDATA

Making narrative-threat intelligence buildable and explainable.

IN PROGRESS
CAPABILITY GAP ANALYSISDETECTION✓BOT ID✗ GAPORIGIN TRACING✗ GAPGEOGRAPHIC SCOPE✗ GAPINTELLIGENCE DEL✓SECTOR AWARENESS✓

RESEARCH TO PRODUCT DIRECTION

Friend / Foe

Designing for confidence when seconds matter.

IN PROGRESS
FIELD RESEARCH // JMRCDECISION LATENCY // OBSERVEDEVIDENCE_01

What I care about

04 PRINCIPLES // METHOD

The questions I return to when building products people need to trust and act on. See the full method →

01

Find the real workflow problem

OBSERVATION > ASSUMPTION

QUESTIONS AT THIS STAGE

  • Where does the work actually break down?
  • What are we assuming that we haven't observed?
  • Who feels the friction most acutely?
02

Build the useful first version

VELOCITY // RELIABILITY

QUESTIONS AT THIS STAGE

  • What is the smallest version that changes someone's day?
  • What must be reliable on day one?
  • What can wait without losing the point?
03

Make complex systems understandable

VISIBILITY // MAPPING

QUESTIONS AT THIS STAGE

  • What is invisible that should be mapped?
  • Where do people lose the thread?
  • What does clear enough to act look like?
04

Design for trust and failure states

SECURITY // HUMAN RESILIENCE

QUESTIONS AT THIS STAGE

  • What happens when this fails?
  • How does someone know they can trust the output?
  • What recovery path exists for the human in the loop?

I like building with people who care about what happens after the feature ships.