Hi,
I'm Angus Chow
AI & Software Engineering Intern @ iGears Technology Β· BSc Computer Science (AI) @ HKBU
I design and ship full-stack products β from AI features and data pipelines to production web apps used by real users.
Tech Stack
π€ AI/ML & Computer Vision
π» Programming Languages
π Web Development
βοΈ Cloud & Infrastructure

π§ Databases & Tools
Work Experience
- Designed and developed an AI-driven dictation practice platform for primary students, architecting a responsive web system and packaging it as a cross-platform mobile application for distribution.
- Built a hybrid OCR & LLM pipeline to scan and structure dictation sources, integrating a VLM and custom text-image alignment algorithms to grade handwritten submissions and corrections.
- Integrated multi-lingual (English/Cantonese) TTS with configurable speed/interval controls, designing adaptive logic to dynamically isolate and replay erroneous segments for targeted, iterative practice.
- Designed user-centric landing pages to optimize conversion rates and managed end-to-end digital display campaigns across SEO/SEM, email, and social media channels.
- Analyzed market trends and campaign performance metrics to optimize marketing spend and drive strategic digital advertising initiatives.
- Prepared channel performance reports and supported campaign content coordination to refine audience targeting and messaging across digital touchpoints.
Education & Other Experience
Hong Kong Baptist University
Sep 2023 β Jun 2027 (anticipated)
BSc Computer Science (AI Concentration) Β· cGPA 3.46 / 4.00
- AIA Scholarship (2023β2027)
- Undergraduate Scholarship in Computer Science (2024)
- Computer Science Department Alumni Scholarship (2025)
French-German Society | Hong Kong Baptist University
Jan 2025 β Jun 2026
Internal Vice President (2024β2025), Financial Secretary (2025β2026)
- Coordinated internal operations and managed organizational documentation to ensure smooth society operations.
- Collaborated with executive committee to plan and deliver cultural and language exchange programs.
Mini-IO | Hong Kong Baptist University
Aug 2024 β Present
Student Member
- Provided peer support and orientation for international students transitioning to university life in Hong Kong.
- Fostered cross-cultural engagement through campus events and student integration initiatives.
Projects
Time Map HK (ζε ε°ε)
- Built an ICTA-entry interactive heritage map with 48+ curated Hong Kong locations, era/district filters, GPS discovery, and bilingual timeline exploration from open government and archive data.
- Architected per-location AI chat with DeepSeek RAG, on-topic guardrails, rate limiting, and streaming responses grounded in verified location context.
- Deployed on Azure (PostgreSQL, Blob Storage, App Service) via GitHub Actions, with a licensed image pipeline sourcing Wikimedia Commons and M+ archives.
Movie Recommendation System
- Architected a Flask recommender with PostgreSQL, comparing Factorization Machine (hybrid CF + content) and SASRec (Transformer-based sequential) models.
- Built user flows for genre preferences, movie feedback, and profile management, with dual UI themes and Google OAuth login.
- Deployed via Docker and Railway; offline evaluation reached P@10 = 0.60 and nDCG@10 = 0.84 for the FM model.
Licence Plate Detection & Recognition
- Designed a modular two-stage pipeline separating plate detection (Faster R-CNN, RetinaNet, DETR) from OCR recognition.
- Implemented CRNN with a shared ResNet-50 backbone and partial freezing to ensure fair cross-model comparison.
- Achieved F1 = 0.85 on 433 images; benchmarked against OpenALPR and identified OCR generalization as the primary bottleneck.
Online Library Management System
- Developed a Vue.js frontend with Composition API, Bootstrap, and Oruga UI for book browsing, borrowing, and account workflows.
- Built an Express.js REST API with JWT authentication and role-based access control for admin and regular users.
- Modeled inventory and borrowing data in MongoDB with Mongoose, including validation, pagination, and secure endpoint design.
Hybrid Zero-DCE Low-Light Enhancement
- Trained a supervised U-Net for paired low-light enhancement on the LOL dataset.
- Built a hybrid ZeroDCE pipeline for zero-reference enhancement without requiring ground-truth image pairs.
- Compared both approaches under a unified evaluation setup to analyze trade-offs in detail recovery and contrast.
Contact
Open to software engineering, AI engineering, and ML internships and roles. Feel free to reach out!