Senior AI Engineer | Computer Vision & Edge AI

Singapore | Portfolio | CV | LinkedIn | Email

I build camera AI at Black Sesame Technologies, connecting portrait matting, depth estimation and rendering to C++/Android deployment and release evaluation.

junxian@portfolio
-----------------
production : computer vision + computational photography
deployment : Python / C++ / Android / TFLite / ONNX / GPU
exploring  : agent orchestration + multimodal workflows
location   : Singapore

• Portrait imaging & computational photography: segmentation, matting, monocular depth and GPU rendering, with boundary analysis and visual-quality evaluation.
• Vision models on edge devices: quantization, export validation, native integration and device regression checks.

Employer work is summarized without customer identities, proprietary code or internal metrics.

• Reliable Agent Workflow: a Python reference for retries, fallback, output validation and human-review gates, with reproducible failure-path tests. Scripted providers; no live LLM integration.
• Photography Mentor: a browser-local image-analysis and practice app with source-linked retrieval. Deterministic diagnostics and coaching rules; Chinese interface.

Engineering notes cover review-gate semantics and auditing the exact Git release snapshot.

Interested in Applied AI and Forward Deployed Engineering, including multimodal systems where model quality, product requirements and runtime constraints meet.