Lokos AI - Delivery OS

Delivery OS is the platform Lokos AI clients use to request, approve, fund and follow software projects. A client organization is onboarded, submits software requests, answers clarifications, signs an AI-generated statement of work and funds it with Stripe, then follows delivery from the client portal.

I am the main front-end engineer and work across the stack: the Next.js web app, the Expo mobile app, the FastAPI delivery modules and the GKE infrastructure that runs them.

Highlights:

1. End-to-end delivery flow: onboarding, agreements, requests, clarifications, SOW signature, Stripe funding and payments, with Stripe and storage provisioned automatically when an organization is activated.
2. Design-system migration: moved the front-end to Next.js and rebuilt it on shadcn/ui and Tailwind, removing Material UI in four incremental phases without freezing feature work.
3. Native mobile client portal: Expo / React Native app with Google and Apple sign-in, Face ID / Touch ID unlock, in-app account deletion and PostHog analytics, shipped through TestFlight and App Store review.
4. AI features: robust SOW generation with LLM schema validation and retries, and earlier work on OpenAI Realtime voice agents with RAG and real-time call metrics over WebSockets.
5. Infrastructure: GKE spot node pools with on-demand fallback, staging environment, registry cleanup and cost reductions, all managed with Terraform.
6. Security: MFA, magic-link member invitations, one-time secret links and dependency CVE patching.
7. Internationalization across web, SMS, voice and prompts (English / Spanish).