Core Features
Selective Listening (tap a face, hear only that person), a Sign Language Bridge for ASL fingerspelling, and Chat Mode with voice-enrolled speaker diarization and a colour-coded live log.
Hello world. I'm
Second-year cybersecurity student at Temple University Japan, based in Tokyo.
01 / about
I'm a second-year Cybersecurity student at Temple University Japan, based in Tokyo. I got into coding out of curiosity and found I genuinely love it. I've worked through Python, C, Java, and web development as part of my degree. My biggest interests right now are cybersecurity and digital privacy, AI and machine learning, and the development and use of various Linux distributions.
I've already had the chance to build real-world projects and place first in hackathons, and I'm eager to keep growing. I'm looking for opportunities to collaborate, contribute, and learn alongside people who hold themselves to a high standard and push each other to be better.
class CarlMasters:
school = "Temple Univ. Japan"
year = "2nd Year Cybersecurity"
skills = ["Python", "C",
"Java", "HTML/CSS/JS"]
focus = ["Cybersecurity & Privacy",
"Operating Systems",
"AI / ML"]
based = "Tokyo, Japan"
def greet(self):
return "Just getting started."
02 / skills
03 / projects
focus-hear-app.vercel.app · SDGs to Startups 2026 · OpenAI x Tokyo AI
A real-time assistive communication platform for people with hearing loss. 430 million people worldwide live with disabling hearing loss, yet every captioning app transcribes everyone at once, flooding users with an unreadable wall of overlapping voices. FocusHear lets you choose whose voice you hear. Built in 48 hours to take first place at SDGs to Startups, then rebuilt from the architecture up and taken to an invite-only Build with OpenAI event, where it advanced to the first round against international teams.
Selective Listening (tap a face, hear only that person), a Sign Language Bridge for ASL fingerspelling, and Chat Mode with voice-enrolled speaker diarization and a colour-coded live log.
Vertical Slice Architecture rebuild on a GPT-4o and ElevenLabs Scribe pipeline, Supabase Auth with Stripe billing, saved voice and face profiles, live caption translation, and a trained ONNX-based LSTM sign-language model replacing the original demo classifier.
Carl Masters, Cassady Mead, Juthathip (Jenny) Loedsinaudom, and Thiago Komeno.
Builders Weekend 2026 — Shibuya, Tokyo, Japan
An AI-powered dining app that removes the chaos of splitting restaurant bills. Guests sign in, provide a voice sample, and the app listens in real time during ordering to identify who ordered each item. When the check arrives, the host scans the bill, costs are assigned automatically, and everyone pays their share hands-free.
Multiple languages and APIs wired into voice and payment workflows, with integrations from Open Network Labs, Lovable, Supercell, VoiceOS, RevenueCat, and ElevenLabs.
Carl Masters, Bhushith Gujjala Hari, Kseniya Chadovich, Kevin Beutler, and Fangyan Fu.
A business website built for a freelance English teacher to advertise their services. Features a clean landing page, a services section, and a contact form. My first project built for a real client.
Designed and coded from scratch to practice modern HTML, CSS, and vanilla JavaScript, with animations, responsive layout, and all. A project that's also the thing showing you the project.
04 / background
Temple University Japan — Tokyo, Japan
Coursework covering programming fundamentals in Python, C, and Java, introductory cybersecurity concepts, and core math for Cybersecurity at Temple University Japan. Learning to think like a security professional from the ground up.
0842 — Field Artillery Radar Operator | Camp Hansen, Okinawa, Japan & Camp Lejeune, North Carolina, USA
Served four years as a Field Artillery Radar Operator, operating the AN/TPS-80 Ground/Air Task Oriented Radar, AN/TPQ-48 Lightweight Counter Mortar Radar, and AN/TPQ-46 Firefinder Radar (now retired). Worked as part of a cohesive team under demanding operational conditions and strict time constraints, contributing to the successful tracking of thousands of artillery munitions and delivering critical battlefield intelligence directly to commanders.
Concurrently held a secondary billet as a Key Management Infrastructure (KMI) Custodian, responsible for the secure handling, safeguarding, loading, and operational use of highly sensitive military cryptographic keys and associated intelligence materials.
Selected to attend two formal military schools, earning qualifications as both a Combat Marksmanship Coach (CMC) and a Combat Marksmanship Trainer (CMT).
Lake Minneola High School — Minneola, Florida, USA
Completed a standard high school curriculum at Lake Minneola High School, building the study habits and work ethic that carry into everything since.
Chiyoda City, Tokyo, Japan
An evening with the Cursor team, hosted by the Cursor Tokyo and Tokyo AI communities at a venue provided by Datadog Japan. Talks from Frances Thai and Nick Miller covered where Cursor is heading and the features behind it, including Cloud Agents, with Nick's “Building Your Own Software Factory” focused on structuring autonomous workflows. A lot of it came down to using the tool deliberately rather than just quickly, which was worth hearing straight from the people building it.
Google Japan, Shibuya Stream — Tokyo, Japan
An evening at Google Japan hosted with Google for Developers, on taking AI systems from prototype to production. A keynote from David McLaughlin, Director of Global Developer Ecosystem at Google, was followed by a Google AI Studio workshop led by Juyeong Ji of Google DeepMind, covering Gemini, Google Cloud, and the architectures behind running AI platforms at scale. A good counterpoint to the local-model session earlier in the month: much the same models, but a very different set of constraints once you are serving them to real users.
Minato City, Tokyo, Japan
A one-day hackathon focused on building production-ready AI agents themed around Tokyo and Japanese culture. I entered this one solo. Every build had to integrate at least one of the partner platforms, so most of the day went into learning tools I had never touched before, spanning GPU infrastructure, model APIs, and agentic coding environments. It was a great opportunity to work with that much new tech in one place, and I appreciated getting to take part.
ai& for full-stack inference on domestic GPU infrastructure, Daytona for secure code execution, GMI Cloud for unified access to 200+ models, Nosana for decentralised GPU compute, Qwen Cloud for reasoning and agentic coding, and Qoder as an agentic coding environment.
Organised by AI Builders, with MeltingHack and Mercari as community partners.
Sakura Deeptech Shibuya — Tokyo, Japan
A single-evening hackathon built around Kimi K2.7 Code, Moonshot's agentic coding model, running on ai&'s inference platform. It was a good opportunity to build and learn on a new inference stack, and to experiment with Chinese open source models that I had not worked with before. A lot picked up in one evening, and a great team to do it with.
Carl Masters, Ruoan Li, Hina ‘Ai’ Sekine, and Thiago Komeno.
Hosted by ai& and Kimi (Moonshot AI) with AMD as co-sponsor, supported by the AI Founders’ Mindset Community and organised by Kaori Rei.
Google Shibuya Offices — Tokyo, Japan
A Vibe Coders Tokyo session hosted at Google's Shibuya offices, with a talk from Juyeong Ji of Google DevRel on running open-weight models locally. The focus was Gemma and similar lightweight models that run on personal machines and private infrastructure rather than through a hosted API, and what that changes for cost, latency, and control over your own data. The tutorial was followed by open build time to get a model running end to end. The privacy case is what stuck with me: a model you run yourself is one nobody else is logging.
Hongo, Tokyo, Japan
An invite-only build event hosted by OpenAI and Tokyo AI. I brought a fully rebuilt FocusHear, replacing the hackathon prototype with a production architecture: a GPT-4o and ElevenLabs Scribe pipeline, authenticated accounts with billing, saved voice and face profiles, and a trained sign-language model in place of the original demo classifier. Advanced to the first round against a strong international field.
See FocusHearShibaura Institute of Technology — Tokyo, Japan
A 48-hour build against UN Sustainable Development Goal 10, Reduced Inequalities. Our team took first place with FocusHear, targeting a gap no captioning app addresses: 430 million people live with disabling hearing loss and none of them can isolate a single voice in a crowded room. Organised by Tokyo Student Network, Tokyo AI, the TUJ CS Society, GDG on Campus at Sophia, IPUT, and Waseda University.
See FocusHearShibuya, Tokyo, Japan
A weekend build sprint with a five-person team from Temple University Japan. We shipped TabeTalk, an AI dining app that identifies who ordered what by voice and splits the check automatically, which meant getting real-time speech recognition and payment workflows talking to each other inside two days. Sponsor integrations included Open Network Labs, Lovable, Supercell, VoiceOS, RevenueCat, and ElevenLabs.
See TabeTalkTemple University Japan — Tokyo, Japan
A design-focused competition where the deliverable was a concept rather than a shipped product. Our team took first place by researching user needs, wireframing, and prototyping a clean interface, then defending every design decision to the judges under a tight time limit. The lasting value was learning to justify a design out loud, not just produce one.
06 / blog
Most resumes never make it past the first 10 seconds. Here's what separates the ones that do: format, phrasing, and the one-page rule.
Read moreYou don't need to be a security expert to protect yourself online. The right habits, consistently applied, cover the vast majority of real-world threats most people face.
Read moreBig Tech apps are convenient, but they come at a cost. These privacy-respecting swaps are just as good, sometimes better.
Read more07 / contact
I'm actively looking for internship and entry-level opportunities in cybersecurity, including SOC analyst roles, penetration testing, security research, and cybersecurity engineering. Whether you have a role, a project, or just want to connect, my inbox is open.