Four curated case studies

The platform required fast MVP delivery, reliable backend services, and scalable architecture decisions as product scope evolved.

• Built and shipped MVP features across authentication, data models, and third-party API integrations.
• Implemented FastAPI backend services with validation and persistence layers for core workflows.
• Led system architecture and backend prioritisation with the Product & Operations Co-Founder.
• Mentored engineers and maintained quality through sprint cycles and review standards.

• Delivered production-ready backend capabilities supporting core platform workflows.
• Improved technical consistency and delivery reliability across sprint cycles.
• Established a scalable technical foundation for continued product growth.

The broker needed to correctly parse and encode Kafka protocol frames while supporting multiple simultaneous client connections.

• Implemented TCP socket binding and binary message parsing for Kafka requests and responses.
• Built goroutine-based concurrent client handling for serial and parallel request processing.
• Delivered DescribeTopicPartitions and ApiVersions APIs end-to-end.
• Started extending support toward Fetch and Produce APIs.

• Delivered a working broker that handles core Kafka API flows with protocol-correct encoding and decoding.
• Validated concurrency approach across multiple simultaneous connections.
• Established a robust foundation for full message consumption and production support.

The system needed reliable orchestration between LLM responses, tool dispatch, and iterative reasoning over multi-step tasks.

• Implemented a core agentic loop for iterative reasoning and response handling.
• Integrated read, write, and bash tools with schema-driven advertisement and execution.
• Built direct LLM API communication and dispatcher logic across tool and model interactions.

• Delivered an autonomous coding agent that can read files, write outputs, and execute shell commands.
• Validated end-to-end orchestration across multi-step coding workflows.
• Created a reusable foundation for extending toolsets and agent capabilities.

The platform needed real-time camera marker detection, persistent scoring, and reliable hosting for live event usage.

• Integrated react-three-arjs for marker detection and 3D rendering.
• Implemented Firestore-backed persistent player scores and live leaderboard logic.
• Delivered frontend, backend, and hosting as a complete full-stack release.

• Enabled event participants to play through AR marker discovery with live score tracking.
• Supported engagement at GDG ANU event through real-time leaderboard visibility.
• Demonstrated full-stack delivery capability for time-bounded event requirements.