Software Engineer
Production-Grade AI Platform
Done Life·Sydney, Australia
Dec 2024 – Present
Synopsis
Leading full-cycle engineering for a production AI-first productivity platform. This spans backend AI pipelines and cloud infrastructure through to App Store compliance and CI/CD, converting unstructured emails, PDFs, and images into structured calendar events for consumers.
The Making Of
🎯 The Challenge
Building a production-grade consumer app from scratch across three fronts simultaneously: (1) AI reliability: LLMs hallucinating dates, poor PDF handling, and prompt drift on real-world email shapes; (2) Platform compliance: an App Store Guideline 3.1.1 rejection requiring an iOS/Android product split; (3) Data security: sensitive item content needed field-level protection without breaking offline Flutter sync or triggering iOS Keychain concurrency crashes.
💡 The Solution
Migrated AI extraction to OpenAI Responses API with strict JSON schema and IANA timezone injection via Luxon, so relative dates resolve correctly in the user's wall-clock zone. Implemented AES-256-GCM field-level encryption with a GCP Secret Manager-wrapped DEK and a serialized async queue to eliminate Riverpod/Keychain race conditions. Split the paywall by platform, with iOS redeeming App Store offer codes and Android retaining the approved voucher flow. Provisioned all GCP resources with Terraform and wired three GitHub Actions workflows for repeatable, secret-safe builds.
🚀 The Impact
App is production-ready on both App Store and Google Play. Field-level encryption protects all sensitive Firestore content without plaintext storage. The CI/CD pipeline enforces dart format, flutter analyze, and flutter test on every PR. Codebase grew ~58% in Dart LOC across 40+ commits in under 3 months, with full infrastructure expressed as versioned code.
Key Contributions
- 1Engineered an AI email ingestion pipeline using OpenAI Responses API with strict JSON schema, native PDF input, and timezone-aware date resolution via Luxon, fully eliminating the temporal hallucination issues that plagued earlier LLM extraction
- 2Implemented AES-256-GCM field-level encryption with server-side DEK wrapping via GCP Secret Manager; built a serialized async queue to prevent iOS Keychain concurrency crashes caused by overlapping Riverpod providers
- 3Provisioned GCP infrastructure as code with Terraform (8 .tf files covering Pub/Sub, IAM, Secret Manager, API services) and authored three GitHub Actions CI/CD workflows for PR validation (format, lint, test) and dev/prod Android AAB builds
- 4Navigated an App Store Guideline 3.1.1 rejection and shipped a platform-split paywall: iOS uses App Store offer code redemption via RevenueCat; Android retains the approved in-app voucher flow
- 5Built multi-auth (Email/Password, Sign in with Apple, Google Sign-In) with platform-specific routing, anonymous-to-linked account flows, and branded transactional email via Gmail API; scaled codebase from ~13,800 to ~21,900 Dart LOC across 40+ commits in under 3 months
Key Metrics
Skills & Technologies
- FlutterMobile Frontend
- Firebase / Cloud FunctionsBackend & Database
- OpenAI Responses APIAI Extraction Engine
- RevenueCatSubscription Management
- TerraformInfrastructure as Code
- GitHub ActionsCI/CD
- GCP Secret ManagerKey Management
- Firebase Analytics (GA4)Product Analytics
Quick Info
- Company
- Done Life
- Location
- Sydney, Australia
- Type
- Founding Role
- Duration
- Ongoing (5+ months)
- Complexity
- Principal