Projects

A self-hosted local media management app built for NAS deployment. Framd indexes your media library, generates previews, and serves them through a lightweight web UI with a digital frame display mode that cycles through your photo library.

For a detailed write-up of the design decisions and progress, see the project blog.

• Reduced full scan time by ~84% (from 12 minutes to under 2 minutes) by building a concurrent indexing pipeline using Java virtual threads and a semaphore-bounded executor, decoupling EXIF extraction, preview generation, and DB persistence using blocking queues
• Designed a fatigue-based frame selection algorithm that groups photos by folder, weights selection by view frequency, and distributes picks across the chronological span of each folder
• Integrated a Java NIO WatchService file watcher to trigger incremental re-indexes on directory changes, avoiding full rescans
• Used content hashing to detect already-indexed files and key preview storage

• Maintain a home lab environment as a personal project, employing enterprise-grade tools and practices
• Design and operate infrastructure using Proxmox and TrueNAS
• Automate system configurations and deployments using Ansible and Docker
• Implement comprehensive backup strategies, security measures and documentation
• Practice DevOps principles through infrastructure-as-code and CI/CD pipelines
• Continuously explore and implement new technologies and security best practices

• Android app that analyses EPUB books using NLP to extract and visualise character relationships, appearances through chapters
• Uses CoreNLP for named entity recognition, with results stored in MongoDB and visualised through interactive D3.js charts and network graphs
• Server-offloading architecture: resource-intensive NLP processing runs in Docker containers on a backend server, keeping the mobile app lightweight

  

• Developed multiplayer video game AI, Network and Logic with a team