Curriculum Vitae
Production vision systems, from model behavior to device release.
I build camera and image AI systems across model development, C++/Android integration, visual-quality evaluation, and production handoff. My public work extends that discipline into reliable agent and multimodal workflows.
Download one-page PDF CV
Profile
Senior AI Engineer with production depth in camera AI, computational photography, edge deployment, and C++/Android integration. I turn model behavior and visual-quality feedback into deployable systems, hard-case evaluation, and maintainable handoff, while building tested public references for reliable agent workflows.
Target roles: Applied AI Engineer, Forward Deployed Engineer, Computer Vision Engineer, and Multimodal AI Engineer.
Professional Experience
Senior AI Engineer | Black Sesame Technologies (Singapore)
Jan 2025 - Present
- Own end-to-end delivery for camera and image AI features, translating product requirements and field feedback into technical scope, prototypes, evaluation gates, and production handoff.
- Lead real-time composition and portrait-imaging systems combining detection, optical-flow tracking, segmentation and matting, monocular depth, GPU rendering, and application state behavior.
- Drive model-to-device delivery across PyTorch, quantization, TFLite/ONNX, C++/Android integration, GPU pipelines, and on-device debugging.
- Coordinate product, camera tuning, runtime, QA, and customer-facing teams around reproducible issue sets, acceptance decisions, and release readiness.
AI Engineer | Black Sesame Technologies (Singapore)
Aug 2022 - Jan 2025
- Productized vision models for portrait segmentation and matting, monocular depth, object detection, and camera-oriented image processing.
- Built repeatable training-to-deployment workflows spanning data preparation, training, quantization, export validation, runtime assumptions, and model replacement checks.
- Integrated models into native C++ and Android camera pipelines with simulation, cross-platform builds, device deployment, and visual debugging.
- Developed depth-aware bokeh and portrait-rendering workflows with hard-case datasets, batch comparison, golden cases, and release-review tooling.
Research Intern | National University of Singapore
Dec 2021 - Jul 2022
- Conducted computer-vision research for construction-site safety, including dataset preparation, image-classification experiments, and spatial risk analysis.
- Co-authored a peer-reviewed Automation in Construction paper on supervised and self-supervised feature representation learning.
Research Assistant | Shandong University
Sep 2019 - Jun 2021
Applied computer vision and machine learning to interdisciplinary psychology research, supporting dataset development, experiments, visualization, and reporting.
Selected Systems
Reliable Agent Workflow
Provider-neutral Python reference for retry, fallback, validated output contracts, confidence gates, traces, and executable hard cases.
Code and tests
Photography Mentor
Browser-local image diagnostics, retrieval, practice planning, review state, and evidence-aware coaching.
Open the system
Production Vision & Edge AI
Training-to-deployment workflows, quantization and export validation, native C++/Android integration, on-device debugging, and release checks.
Read selected system evidence
Technical Stack
- AI and vision: PyTorch, TensorFlow, OpenCV, segmentation, matting, monocular depth, detection, optical flow, local retrieval.
- Runtime and deployment: Python, C++, Android/NDK, TFLite, ONNX, ONNX Runtime, TensorRT, OpenGL/GLSL, Docker, FastAPI/Flask.
- System quality: quantization, export validation, validated output contracts, retry and fallback references, hard-case suites, traces, release gates.
Education & Publication
Master of Technology, Intelligent Systems | NUS-ISS
2021 - 2022 | GPA 4.33 / 5.00
Bachelor of Engineering, Software Engineering | Shandong University
2017 - 2021 | GPA 86.86 / 100 | Outstanding Graduate
Peer-reviewed publication
Co-author, "Automated classification of 'cluttered' construction housekeeping images through supervised and self-supervised feature representation learning," Automation in Construction, 2023. DOI