Preksha Shah
Book a call

Case studies

Work built to test a real technical idea.

Each one is a real build, with a specific technical bet at the center of it. More are on the way.

Faultline's review summary: prepared by Preksha Shah, target a multi-tenant RAG application, MIT licensed, reviewed by reading the source, one critical and three high findings of six

Faultline

A real production-readiness review of a live multi-tenant RAG app: a critical cross-tenant leak, six ranked findings, and a 90-day plan, the literal deliverable of the Architecture & AI Review offer.

Automated and manual code review, zero live compute; report in TypeScript, React, Vite, Tailwind

Logbook's public homepage, with the pitch: most technical debt isn't tracked, it's just remembered by whoever took the shortcut

Logbook

A decision and tech-debt registry built around a single idea: reversibility, not a backlog label, is what should decide how much scrutiny a decision gets.

Next.js, TypeScript, Zustand, Supabase, Vitest

Runway's public landing page, with the pitch: how long until the money runs out, and what actually changes it

Runway

A cash-runway calculator built around one question: are you default alive or default dead, and how much time do you actually have.

Next.js, TypeScript, Zustand, Supabase, Vitest

Triage's report overview: the one-number summary showing the model missed 9 of 18 escalations and the code caught 7 of them, alongside the four things that separate the build from a demo

Triage

A support-ticket triage agent built to prove the distance between an AI demo and an AI system: a real eval run with published numbers, a measured cost ceiling, and tested prompt-injection defense.

TypeScript, Node.js, Groq, Gemini, Zod, Vitest, React, Vite, Tailwind