Wayne Fourie's Work | ContraWork by Wayne Fourie
Wayne Fourie

Wayne Fourie

Senior AI & Software Engineer | LLM/RAG, Computer Vision

New to Contra

Wayne is ready for their next project!

Cape Town, South Africa

WorkPostsServicesAbout

Cape Town, South Africa

Cover image for CamBridge - Camera Middleware Engineered a
CamBridge - Camera Middleware Engineered a protocol translation layer that bridges Chinese IP cameras (Hikvision, Dahua) with professional NVR platforms like Milestone and Genetec. The middleware abstracts vendor-specific protocols into a unified interface for stream management and event processing. Key features: - Protocol handlers for Hikvision ISAPI, Dahua, and generic ONVIF - Real-time event subscription (motion, line crossing, face detection) - RTSP stream discovery and proxy configuration - 24/7 health monitoring with automatic stream loss detection Tech: Python, asyncio, aiohttp, RTSP, ONVIF, ISAPI, FFmpeg
0
16
Cover image for Agentic RAG Pipeline Created an
Agentic RAG Pipeline Created an intelligent email triage system that classifies incoming emails and generates context-aware draft responses using retrieval-augmented generation. The LangChain agent uses tools to search a knowledge base, check customer history, and apply escalation rules. Key features: - 7-category classification with 4 priority levels - Tool-augmented LLM agent for dynamic information retrieval - Vector knowledge base storing policies, FAQs, and response templates - Structured output with confidence scores and routing recommendations Tech: Python, LangChain, GPT-4o, ChromaDB, OpenAI Embeddings, Pydantic
0
44
Cover image for Trademark Similarity Engine Developed a
Trademark Similarity Engine Developed a visual similarity search system for trademark infringement detection using CLIP embeddings and PostgreSQL vector search. The system encodes trademark images into 512-dimensional semantic vectors and finds visually similar marks in milliseconds. Key features: - 87-94% accuracy on trademark similarity detection - Sub-50ms query latency using pgvector IVFFlat indexing - Scales to 1M+ trademarks with approximate nearest neighbor search - Configurable similarity thresholds with automatic infringement flagging Tech: Python, CLIP (ViT-B/32), pgvector, PostgreSQL, FastAPI, PyTorch, GCP
0
46
Cover image for Built an intelligent document extraction
Built an intelligent document extraction engine that automatically parses invoices using GPT-4 Vision. The system extracts 15+ field types (vendor info, line items, totals, dates) with 93.5% accuracy and includes confidence scoring for each field. Key features: - Multi-modal LLM processing for PDF, images, and scanned documents - Cross-validation logic that catches extraction errors by comparing line totals vs subtotals - Async batch processing for high-volume workflows - Pydantic schemas for type-safe, structured output Tech: Python, GPT-4 Vision, LangChain, Pydantic, FastAPI
0
54
Cover image for Built an intelligent document extraction
Built an intelligent document extraction engine that automatically parses invoices using GPT-4 Vision. The system extracts 15+ field types (vendor info, line items, totals, dates) with 93.5% accuracy and includes confidence scoring for each field. Key features: - Multi-modal LLM processing for PDF, images, and scanned documents - Cross-validation logic that catches extraction errors by comparing line totals vs subtotals - Async batch processing for high-volume workflows - Pydantic schemas for type-safe, structured output Tech: Python, GPT-4 Vision, LangChain, Pydantic, FastAPI
0
54