> Yong Zane_
// Background
Architecting scalable AI systems that transform research into production
// Work Experience
Research Engineer
Key Achievements:
- ▸Architected and deployed a highly available LangGraph-based multi-agent system for dynamic task decomposition in complex manufacturing pipelines, resulting in a 50% reduction in decision-making latency and enabling scalable workflow automation.
- ▸Fine-tuned state-of-the-art Vision-Language Models (VLMs) and object detection models leveraging both Transformer and CNN architectures for structured document parsing of 2D engineering drawings, increasing data ingestion throughput by 40%.
- ▸Engineered and productionized highly scalable backend microservices using FastAPI and ASP.NET, containerized via Docker, and orchestrated with Kubernetes, supporting critical operations and achieving 99.9% uptime.
- ▸Developed and shipped a multi-tenant Angular platform utilized by 10+ internal/external manufacturing teams for the real-time deployment, monitoring, and control of AI-powered workflows.
- ▸Integrated and enforced MLOps standards using MLflow for experiment tracking and inference monitoring across pipelines, reducing model deployment time from 2 days to 4 hours.
Technologies Used:
Machine Learning and Image Processing Research Assistant
Key Achievements:
- ▸Improved imaging sensitivity by 90% for single-particle tracking (SPT) through the implementation of unsupervised learning image denoising.
- ▸Built a foundational large-scale microscopy imaging dataset and trained a regression model to enable accurate and automated protein size prediction.
Technologies Used:
Optical Engineer Intern
Key Achievements:
- ▸Implemented image processing algorithms using OpenCV library to assess the optical performance of the instrument with a 25% acceleration of the feature extraction process while maintaining a 95% accuracy.
- ▸Applied scikit-learn regression to model fluorescence dye behaviour across light spectra, achieving 80% prediction accuracy.
Technologies Used:
// Education
// Publications
// Projects
Surgical-Tray Optimizer (STOp)
NTU CAO x ST Engineering Hackathon 2024
▸Introduced an optimization model that efficiently configures the required surgical tray for cataract surgery, resulting in a cost reduction of $4 million annually.
▸Conceived a solution to utilize predictive analytics for assessing the risk of adverse events during operations, reducing the possibility of complications during surgery by 50%.
▸Secured first runner-up position among 11 participating teams from a wide variety of disciplines at NTU.
Video-Captioning Filtering Pipeline
NTU EEE Machine Learning and Data Analytics (MLDA) - Deep Learning Week 2024 Hackathon
▸Designed a power-efficient Python-based filtering pipeline to select high-quality video-caption pairs, improving the quality of the text-to-video generative model by 12%.
▸Deployed machine learning based evaluation metrics (PyTorch) that reduce carbon emissions generated by 50% during the model training process.
▸Achieved second place out of over 50 participating teams from various educational institutions in Singapore.
ConnectEZ - AAC Device for Aphasia Patients
Medical Device Design and Biomedical Project Design & Management
▸Led a team of 8 individuals in the development of an affordable AAC device, resulting in enhanced communication capabilities for Aphasia patients and a positive impact on their mental well-being.
▸Utilized Google Firebase Realtime Database to securely log sensor data, enabling personalized treatment solutions while prioritizing user privacy and data security.
▸Provided technical guidance to the team in proficiently utilizing programming languages including C, C++, and Arduino IDE to develop an advanced algorithm for user fall detection.