Gnangze Elie Gildas Traore

AI Engineer

Industrial Technology Research Institute (ITRI) · Hsinchu, Taiwan

Applied research institute. UAV computer vision and vision-language model (VLM) workflows, from dataset curation to edge deployment.

• Built and shipped UAV-focused computer vision and VLM workflows end to end, from dataset curation to model evaluation and deployment.
• Developed an automated long-tail synthetic data pipeline that generates, validates, and version-tracks augmented samples with a gated, verification-first workflow.
• Deployed real-time inference services via APIs for internal integration, enabling repeatable evaluation runs and production-style testing.

Lead Engineer (contract)

Business platform for a consulting firm (independent contract) · Remote

Full delivery of a multi-module business platform for a paying client, won through a competitive request for proposals. Client details confidential.

• Designed and built a NestJS + Next.js modular monolith (PostgreSQL, Redis, Prisma) with a plugin-style payment hub integrating mobile money and card aggregators.
• Implemented granular role-based access control, mandatory two-factor authentication, and a bilingual English/French product with mirrored i18n keys enforced in CI.
• Set up production operations: Traefik/Caddy routing, Grafana/Prometheus/Loki observability with alerting, WAL-streamed database backups to object storage with rehearsed restore drills.

Founder, AI Systems Engineer

CaveMind Labs · Hsinchu, Taiwan

Own venture: production-grade agentic AI and computer vision services.

• Built a self-hosted Private Agent Runtime that turns LLM workflows into reliable software: durable state, restart-safe execution, drift-resistant scheduling, real-time streaming via Server-Sent Events, notification routing, and a human acknowledgement (ACK) loop with a full audit trail.
• Integrated multi-model support (OpenAI and Gemini) behind a model-agnostic layer; used LLM semantic analysis to filter noise, summarize meaningful changes, and classify severity for monitoring use cases.
• Architected, developed, and deployed a service-oriented image and video restoration application serving multiple Vision Transformer (Uformer) models with VRAM-aware resource management and an intelligent cache (FastAPI backend, Next.js/React frontend, WebSockets, Docker images published on Docker Hub).

AI Engineer

Juxin Digital Technology Co., Ltd. · Taiwan

• Developed and trained Vision Transformers (ViTs) for advanced low-light image and video enhancement.
• Optimized models for next-generation, low-latency performance on production hardware.
• Integrated the enhancement model into a multi-stage vision pipeline feeding a downstream instance segmentation model.

Full-Stack Developer and AI Engineer

InsightMed Diagnostics · Hsinchu, Taiwan

• Independently architected and developed a full-stack ECG (electrocardiogram) labeling web application with real-time visualization, annotation, database management, and Docker containerization for reproducible deployment.
• Engineered the backend with Django and Django Channels for WebSocket-based real-time communication; built the interactive front end with Dash (Plotly), HTML, and JavaScript for manual annotation of P-waves, QRS complexes, and other key features.
• Implemented deep-learning-based automated time-series labeling with continual learning from domain-expert feedback.

National Yang Ming Chiao Tung University (NYCU)

Master of Science (MSc), Electrical Engineering and Computer Science

GPA 4.22 / 4.30

Courses: Advanced Machine Learning, Advances in 3D Vision

Tamkang University (TKU)

Bachelor's degree (with honors)

Fu Jen Catholic University, Chinese Language Center

Language program, Mandarin Chinese