// SPECIALIZATION:Cybersecurity Analyst
and AI Engineer
> NAME: THIEN QUY PHAM
> CORE SKILLS: Cybersecurity, Threat Defense, AI, Python, SQL.
> LATEST PROJECT: Sentra — a team-built AI phishing detection platform that blocks prompt-injection attacks before they reach the LLM.
System Profile
Thien Quy
ACTIVELY SEEKING INTERNSHIPS
Technical Profile
Core CompetenciesSys_Online
Current Role
IT ADMIN II
Systems Managed
1,000+ ENDPOINTS
Focus Area
AI CYBER DEFENSE
Core Stack
PYTHON / SQL / INTUNE
Overview
> BACKGROUND: I am a Computer Science student specializing in Cybersecurity, Artificial Intelligence, and IT Administration.
> EXPERTISE: I build secure systems and leverage AI models for proactive threat detection. My experience ranges from managing enterprise endpoint networks to designing intelligent tools that prevent digital exploits.
> OBJECTIVE: Seeking a Summer 2027 Cybersecurity internship. I am eager to apply my technical skills to incident response, network defense, and scalable security architecture.
Cybersecurity Certificate
VerifiedIssued by: Google / Coursera
Completed hands-on training in threat detection, network security, SIEM tools, and incident response. Prepared for defensive engineering and SOC operations.
CompTIA Security+
PursuingTargeting: Threats, Security Architecture & Operations
Actively deepening knowledge in threat detection, incident response, and securing enterprise infrastructure.
Skills & Technologies
SELECT ICON FOR DETAILSRELEVANT Experience
ORGANIZATION: University of Toledo - Dept. Dean-Engineering
IT System Admin II
- ■Manage 1,000+ Windows and macOS endpoints via Active Directory and Microsoft Intune, enforcing device compliance policies and triaging security alerts across the fleet.
- ■Responded to 15+ security incidents in four months: remediated unpatched Windows 10 endpoints carrying open vulnerabilities and isolated 2 malware-infected devices from the network.
- ■Own endpoint lifecycle management: imaging, deployment, OS and application patching, and hardware refresh cycles that keep the fleet in compliance.
- ■Provide Tier 2 support for 250+ classrooms and offices, resolving desktop, network, printer, and peripheral failures and documenting root cause for the Tier 1 queue.
ORGANIZATION: University of Toledo - Toledo, OH
Undergrad Research Assistant
- ■Develop Python/OpenCV video tracking and behavior-state analysis to extract zebrafish motion features from Novel Tank Test footage.
- ■Prepare labeled trial data and training workflows that estimate compound-family probabilities, evaluate models, and flag novel responses.
- ■Bridges computer science with clinical research, held concurrently with the IT System Admin role.
ORGANIZATION: University of Toledo - IT
IT Support Technician
- ■Triaged 10-15 tickets per week through UToledoHelp, spanning same-day fixes to multi-day network and hardware escalations.
- ■Escalated each case through a structured process, writing resolution notes clear enough for the next technician to reproduce the fix.
- ■Held concurrently with the IT System Admin and Research Assistant roles.
ORGANIZATION: Harbor Freight
Sales Associate
- ■Maintained email capture above 80% per transaction and membership sign-ups at 1.5%, meeting store growth goals.
- ■Kept extended warranty sales above 11.5% and consistently hit 90%+ of daily sales budget.
- ■Handled data entry and floor work simultaneously in a high-traffic store, adapting quickly to changing needs using judgment and problem-solving.
- ■Explained technical product details in simple, clear steps customers could easily follow.
ORGANIZATION: Explorers Academy
Summer Camp Tutor
- ■Instructed students in computer science fundamentals, teaching algorithms, logical reasoning, and hands-on programming.
- ■Mentored projects focused on STEM creativity, breaking down abstract technical concepts for non-technical audiences.
- ■Developed strong presentation and technical communication skills applicable to cybersecurity user education.
ORGANIZATION: Best Buy
Sales Advisor
- ■Bridged customer needs with appropriate technology solutions through active listening and deep product knowledge.
- ■Maintained continuous learning on new tech hardware and software to provide accurate, informed recommendations.
- ■Cultivated consultative communication skills to quickly build trust with non-technical stakeholders.
Featured ProjectsRestricted
AI Phishing Detection Platform
> Sentra - a team-built AI phishing detection platform for Gmail/Outlook. Engineered a regex pre-flight filter that blocks jailbreak and prompt-injection attempts before they reach the LLM, cutting blocked-request response time from 15-20s to under 5ms.
AEGIS
> Autonomous Agent Security Guard & Prompt Injection Firewall (MakeUC 2026). I designed the firewall workflow, built the admin and intelligence workers that process firewall events and draft new rule candidates, and redesigned the admin and classroom consoles. In end-to-end testing, an approved AI-drafted rule blocked the same attack in a new session.
Port Vulnerabilities Scanner
> Terminal-based Python vulnerability scanner to identify potential network risks, featuring multithreading & SQLite logging.
Academic Background
The University of Toledo
B.S. Computer Science
Focus: Cybersecurity & AI
Academic Standing
Current Coursework (In Progress)
Completed Classes
System Architecture
This portfolio features a Retrieval-Augmented Generation (RAG) pipeline built from scratch. Click through the architecture nodes below to inspect the simulated telemetry and data flow.
Data Ingestion & Embeddings
Personal resume, projects, and skills data are processed and converted into dense vector embeddings using high-dimensional embedding models, mapping semantic meaning into mathematical space.
Vector Search & Retrieval
When a user asks a question, the system computes the cosine similarity between the query embedding and the stored document vectors to retrieve the most relevant context dynamically.
Next.js Edge API
A fast, serverless route acts as the orchestration layer, handling the incoming query, performing the vector math operations locally, and constructing an augmented prompt.
LLM Generation
The augmented prompt containing the retrieved context is sent to a Large Language Model (LLM), which generates a highly accurate, context-aware answer grounded in my actual background.
Secure Comms
STATUS: RELAY NODES ONLINE