Julius Lin

About Julius Lin

30 year of post-PC era/Ubiquitous computing, 3D CV (Computer Vision) AI/ML, and HMI (virtual devices) development.• Over 25 years of IS & T experience specializing in data warehouse, math, and statistics models. • 10 years of extensive database/cluster theory, practice(Oracle/PostgreS), data model and design • 15 years of Python/C++ and shell/Matlab/R scripting and programming and admin. Linux & Windows

Experience

1. Software Architect

   Aitek.Com

   May 2022 — Present

   Major co-author of Nature-BioSensing (provided ~90% of content) for the paper \"Non-invasive Jaundice Detection using Spectral-Band Expansion from RGB Images and Direct Hyperspectral Images\" at is a real-time phone-/Hyperspectral Imaging (HSI)-based jaundice detection system using RGB/multi-spectrum paired with HSI. JaundiceAI-Mobile achieved R² = 0.988, Pearson r = 0.9945 on patient images from NTUH Hospital.I Built dual-norm pipelines across lighting types and expanded.heic/jpg/dng RGB images into 13-channels, enabling pattern learning consistent with HSI Z-profiles. Generated raw Z-profiles from dynamic interferograms captured by PARC\'s FTIR camera (λ = 400–1100 nm, 200 frames/stack) by tracking sclera and palm center/creases using AI models, followed by a custom Fourier Transform to get Z-profile. Applied super-resolution on mobile to simulate hyperspectral output.Created Python/QT5-based diagnostic tools for RoI normalization, anomaly detection, and deep learning model training (SMOTE, DCNN with Attention) to accurately predict jaundice index (mg/dL).2. Airway Management AI: an AI-assisted laryngoscope intubation guidance system. Managed CVAT video annotation under Docker/WSL with hierarchical segmentation labels including physical and virtual super-/sub-labels. Converted \'CVAT for video 1.1\' exports into YOLOv8 Segmentation 1.0 format, resolving missing semi-auto labels via OpenCV:TrackerMIL. Trained a YOLOv11 model TCN using multi-camera inputs (VL and Stylet) for real-time tracking and tube depth estimation during navigation.3. Algorithmic Trading Bot: a multithreaded, rule-based quantitative trading engine integrated with Yahoo and Schwab APIs/data. Developed hybrid scalping and trending across various trading modes. Implemented adaptive signal generation using indicators SMI/MACD, and designed dynamic channel regression with guardrail logic for inter-/intra-day trading and backtest.

Education

• Cleveland State University

  Studied in DBA (Doctor in Business Administration) program

  1993 — 1994

• Nation Taiwan University

  Bachelor’s Degree, Forestry (Industry)

  1985

• Cleveland State University

  Master of Science (MS), Computer Science

  1989 — 1992