AI Engineer | Cloud-Native MLOps | Final-year Robot-AI @ UEH

I build robust AI systems—from training deep learning models to architecting scalable, real-time inference pipelines on the cloud.

Cloud-Native Inference Engine for Real-Time PPE Compliance

• Architecture: FastAPI inference engine with async event logging via ThreadPoolExecutor. (Kafka integration: roadmap)
• AI Pipeline: Dual-YOLO spatial-temporal ensemble + ByteTrack.
• Highlight: Engineered a Forward-CAM interceptor on EfficientNetV2-B0 to interpret feature extraction and suppress False Positives.
• Database: Idempotent event logging to Supabase (PostgreSQL) with local blob storage.

Automated GitOps Architecture & Data Drift Observability

• Infra as Code: Provisioned AWS EC2 infrastructure autonomously using Terraform.
• Observability: Engineered Population Stability Index (PSI) monitoring with Telegram API Webhooks for real-time Covariate Shift alerts.
• Continuous Training (CT): Automated model retraining and MLflow registry updates via GitHub Actions.

Graduation Internship Project

• Integrating IoT sensor arrays (DHT22, CCS811) with MQTT and HomeAssistant.
• Time-series forecasting with LSTM and visualizing telemetry in a Unity 3D interactive dashboard.

• AI / Computer Vision: PyTorch YOLO timm OpenCV ONNX Runtime
• MLOps / Cloud: Docker Terraform AWS (EC2, S3) MLflow GitHub Actions
• Backend / Data: Python FastAPI Kafka Supabase MQTT

• Deep Learning Specialization — DeepLearning.AI
• MLOps Specialization — Duke University