
Miira AI
Software Developer
Canberra, ACT
- Architected a real-time e-commerce recommendation system using Django, PostgreSQL, and Redis, implementing a multi-tier caching strategy (instance, process, and Redis layers) that reduced API response times from 500ms to 50–100ms and achieved 90–95% cache hit rates for personalized product recommendations.
- Engineered a binary tree-based interactive recommendation engine with ADAPT (Adaptive Personalization Through Feature Weighting) and integrated multiple ML algorithms (Collaborative Filtering, Content-Based, BPR, FWLS, LDA), processing real-time user interactions with sub-100ms latency for Shopify stores with up to 1,000 products.
- Developed a production-ready Shopify embedded app with Remix, PostgreSQL (Prisma), and React (Polaris), implementing event sourcing for analytics, session management with identity merging, and webhook-based order attribution, serving real-time product recommendations via API proxy.
- Engineered a zero-dependency storefront widget using TypeScript and ESBuild, delivering gesture-based product discovery with under 300ms interaction latency, integrated with a machine learning recommendation service for personalized shopping experiences.
- Developed a FastAPI microservice leveraging Fashion-CLIP and BLIP2 models to automatically extract 500+ fashion attributes from product images, with direct PostgreSQL integration generating one-hot encoded features that improved recommendation system query performance by 10x.

Ingroe
Software Developer
Canberra, ACT
- Built a RAG-based AI assistant platform using Next.js and TypeScript, featuring real-time chat with streaming responses, voice transcription using OpenAI Whisper, text-to-speech capabilities, and document management with AWS S3 integration for knowledge base storage.
- Fixed S3 file retrieval failures caused by URL-encoded file paths by implementing proper URL decoding in the parsing logic, enabling successful document content extraction from AWS S3 buckets for handwritten document processing.
- Resolved chat interface crashes when the backend API returned malformed JSON responses by adding error handling that catches parsing errors and displays user-friendly fallback messages, improving reliability during document query failures.
- Integrated NextAuth authentication supporting both email/password credentials and Google OAuth, handling different user ID sources across providers by implementing custom JWT callbacks that normalize user identification in session management.

Shine Solutions Group
Software Intern
Canberra, ACT
- Built a serverless ETL pipeline on AWS Step Functions that validates and transforms large datasets through parallel Lambda functions prioritising query performance and storage optimisations.
- Developed a CI/CD pipeline with Terraform modules to automate cloud resource deployments across multiple development environments, enforcing configuration validation to improve deployment reliability and integrity.

Mazaal AI
Software Intern
Sydney, NSW
- Developed modular data pipelines with Python and Kedro framework to analyse multi-format datasets, drive predictive analytics, and deliver actionable insights for decision-making.

Australian National University
Master of Computing

Macquarie University
Bachelor of Applied Finance
