🟢 Open to work
I'm on a journey to enhance my skills and explore new career paths in technology and marketing, ready to make an impact.
Dive into my hands-on projects and externships, where I analyze data and develop strategies to enhance business performance.
Breaking Games · Jun 2026 · ✅ Verified by Extern
A Breaking Games externship focused on analyzing real e-commerce marketing and sales data to understand online purchase behavior and conversion funnels.
.jpg)
Hydroficient · Aug 2026 · ✅ Verified by Extern
Build, hack, and defend real IoT infrastructure, then hand your security playbook to an actual company. You’ll launch realistic cyberattacks, from spoofing and replay attacks to unauthorized shutoffs, and implement defenses like TLS and mTLS.
Engineered a multi-agent AI system using Claude to automate the end-to-end job application workflow. Developed specialized agents for job matching, resume tailoring, ATS scoring, and recruiter review. Integrated public ATS scoring APIs and iterative AI feedback loops to optimize resume quality. Auto
Developed the foundational cybersecurity knowledge including OSI layer, Brute Force detector , exploitation frameworks, web application security, digital forensics.
Developed a responsive e-commerce site with dynamic search, cart, and checkout features. Collaborated in an Agile workflow to plan, debug, and deploy updates efficiently.
.jpg)
Pfizer · ✅ Verified by Extern · ⏱️ In progress
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
I'm on a journey to enhance my skills and explore new career paths in technology and marketing, ready to make an impact.
Cybersecurity fellow with a Bachelor's degree in Economics, eager to strengthen my resume and enhance my job competitiveness. Currently, I'm exploring opportunities in the cybersecurity, marketing analyst sectors to build a solid foundation for my career.
Externships
Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship
Pfizer
Experience
Cantonese Instructor
Hills Learning - Remote · Feb 2023 – Present
Hydroficient IoT Cybersecurity Extern
Extern · Jun 2026 - Aug 2026
E-Commerce Data Analysis & Strategy Externship
Breaking Games · Mar 2026 – Jun 2026
Relationship Banker
East West Bank | Cathay Bank – Brooklyn, NY · Apr 2022 – Oct 2023
Accounting Teacher
Chan Sui Ki Perpetual Help College · Sep 2019 - Aug 2020
Education
The Knowledge House
Innovation fellowship: Cybersecurity track
UDACITY
AWS AI Practitioner Challenge Certificate · Class of 2026
NPower
Application Development Program · Class of 2025
Syracuse University
Certificate of Advanced Study in Data Science · Class of 2024
Hunter College (CUNY)
Bachelor degree in Economics, Minor in Mathematics · Class of 2013
Skills
✅ Verified by Extern
A Breaking Games externship focused on analyzing real e-commerce marketing and sales data to understand online purchase behavior and conversion funnels.
The work analyzed marketing and sales datasets from Google Analytics, Meta Ads, and Shopify to map customer journeys and surface conversion issues. Deliverables included cleaned exports, charts comparing ad metrics, a Claude-generated funnel dashboard, and notes on data-cleaning needs. The work was awarded Top Performer by Breaking Games.
I produced cleaned data exports and comparative visualizations (ad metrics, traffic sources, Shopify sales), iterated Claude prompts to generate a consolidated e-commerce funnel dashboard, and documented data-cleaning steps and insights for future analysis.
✅ Verified by Extern
Build, hack, and defend real IoT infrastructure, then hand your security playbook to an actual company. You’ll launch realistic cyberattacks, from spoofing and replay attacks to unauthorized shutoffs, and implement defenses like TLS and mTLS.
The work examined a 500-room hotel's water management IoT system, producing threat models, an instrumented MQTT pipeline, and a security assessment that measured defenses. Deliverables included certificates and provisioning policy, replay-attack test results, a real-time dashboard, and an integrated Isolation Forest. The work was awarded Top Performer by Hydroficient.
I delivered a professional threat model with top findings, a runnable Python sensor log generator, an instrumented MQTT topic map, TLS/mTLS configuration with measured latency, a Device Provisioning Policy, replay-attack test results, a slide deck and PDF dashboard, and an integrated Isolation Forest anomaly detector.
Engineered a multi-agent AI system using Claude to automate the end-to-end job application workflow. Developed specialized agents for job matching, resume tailoring, ATS scoring, and recruiter review. Integrated public ATS scoring APIs and iterative AI feedback loops to optimize resume quality. Auto
Developed the foundational cybersecurity knowledge including OSI layer, Brute Force detector , exploitation frameworks, web application security, digital forensics.
Developed a responsive e-commerce site with dynamic search, cart, and checkout features. Collaborated in an Agile workflow to plan, debug, and deploy updates efficiently.
✅ Verified by Extern · ⏱️ In progress
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
The externship developed a prototype pipeline that combined OCR, LLMs, and RAG methods to extract and synthesize information from enterprise PDFs. The work included researching LLM fundamentals, designing how models process and generate text, and documenting tradeoffs between model types and prompting strategies.
I produced written analysis explaining LLM tokenization, vectorization, Transformer self-attention, next-token generation, and the roles of pretraining, fine-tuning, and human-preference training, including noted model failure modes.
I explained how LLMs tokenized text into vectors, used the Transformer self-attention mechanism to model context, and produced outputs by predicting next tokens. I described pretraining on large corpora, followed by fine-tuning and human-preference training, and noted that models can still err.
