Muhamad
Amirul
Aiman

Automated defect detection system for Hicom Diecasting — pairing a custom YOLOv8 model with a FastAPI backend and a physical conveyor assembly for real-time pass/fail classification.

• Custom YOLOv8 model trained for high-confidence defect detection on die-cast components
• Sub-200ms hardware-software handshake — zero missed detections under continuous load
• Automated Supabase data pipeline with 100% logging accuracy, eliminating manual entry
• Live Streamlit dashboard for operator oversight and real-time inspection metrics

Serverless backend for a generative AI solution built at VIBE 2026 — orchestrating large-model inference through Amazon Bedrock via Lambda and API Gateway.

• Amazon Bedrock (Nova Pro) orchestration for structured AI output across multi-step workflows
• AWS Lambda execution paths handling inference triggers without persistent server overhead
• Prompt engineering and tokenization optimization to cut latency within AWS cost limits
• Auto-evaluation platforms generating Vite + React frontends via AI prompting

Comprehensive pharmaceutical platform digitizing inventory control, transaction processing, and real-time stock tracking for a simulated drugstore.

• PHP + MySQL stack managing 20+ SKUs, reducing manual audit time by 40%
• Secure checkout module processing transactions with 100% financial accuracy
• Responsive filter interface increasing product retrieval speed by 25%

Desktop library management tool handling 150+ book records, member profiles, and lending logic with zero-corruption file persistence.

• Encapsulation and Inheritance applied to model lending rules and book/member hierarchies
• Automated fine-calculation algorithm improving enforcement consistency by 20%
• Custom error-handling suite reduced system crashes by 15% in final testing