Academic Portfolio of a Visual AI Engineer: From Video Restoration to Multimodal Systems

This article introduces the open-source academic portfolio website of ZhangWeixun, a senior machine learning engineer. It showcases his technical experience and expertise in production-grade visual AI, covering cutting-edge directions like video super-resolution, HDR imaging, LoRA model adaptation, and multimodal AI systems. The website is designed in a minimalist academic style and hosted on GitHub Pages, serving as an excellent example for technical practitioners to showcase their professional capabilities.

• Author/Maintainer: ZhangWeixun
• Source Platform: GitHub
• Original Title: ZhangWeixun.github.io
• Original Link: https://github.com/ZhangWeixun/ZhangWeixun.github.io
• Publication Date: 2026-06-07

This portfolio is ZhangWeixun's personal academic showcase platform. As a senior engineer specializing in computer vision and machine learning, he presents his technical experience in production-grade visual AI through this project. The website uses a black-and-white minimalist academic style and covers multiple cutting-edge technical directions.

Video Super-Resolution and Restoration

Covers deep learning model optimization for video frame information utilization, as well as intelligent video restoration based on Visual Language Models (VLM), applicable to scenarios like post-production and old film restoration.

HDR Imaging and Color Grading

Studies intelligent SDR-to-HDR conversion and AI automatic color adjustment technologies, applied in fields like streaming media and film production.

LoRA Base Model Adaptation

Explores efficient fine-tuning methods for adapting pre-trained large models using low-rank matrices, reducing computational costs and enabling strong AI models to run on consumer-grade hardware.

Multimodal AI Systems

Builds intelligent systems for cross-modal understanding and generation, with great potential in scenarios like content understanding and intelligent recommendation.

Pure Frontend Technical Solution

Implemented using HTML, CSS, and JavaScript without build tools or frameworks, demonstrating in-depth understanding of Web standards, low maintenance costs, and fast loading speeds.

GitHub Pages Hosting

Hosted using GitHub Pages, seamlessly integrated with Git repositories, facilitating content updates and version management.

Responsive Design

Adapts to desktop and mobile devices, ensuring a good browsing experience across different devices.

Interactive Features

Implements interactions like image comparison sliders and mobile navigation via JavaScript to enhance user experience.

Homepage (index.html)

Includes core sections like personal profile, skill showcase, project list, educational background, and contact information, with a clear structure.

Life Page (life.html)

Displays personal photos and hobbies, adding a human touch.

Project Detail Pages

• super-resolution.html: 4K Super-Resolution and VLM Video Restoration
• narrowband-hd.html: Narrowband HD Enhancement
• color-grading-hdr.html: Color Grading and HDR Remastering Each page includes detailed technical explanations and effect comparisons.

Example of an Academic Job Application Portfolio

Provides a concise, professional, and clear reference for machine learning practitioners to create job application portfolios.

Technical Direction Learning Guide

The showcased video super-resolution, HDR, LoRA, and multimodal AI are all hot topics in the field, serving as references for learning path planning.

Pure Frontend Project Practice

Demonstrates how to build a fully functional modern website using basic Web technologies, suitable for frontend learners as a reference.

ZhangWeixun's portfolio not only showcases his visual AI technical experience but also reflects excellent engineering practices and design concepts. Maintaining a high-quality personal portfolio is an effective way for technical practitioners to showcase their capabilities and build a personal brand. This project proves that solid basic technologies can also create impressive showcase platforms. As visual AI develops, engineers with production-grade system development experience will be more valuable, and this open-source project provides a valuable learning resource for the community.