This profile repository currently does not declare a repository-wide software license. Copyright therefore remains with the author by default. Obtain permission before copying, modifying, or redistributing the generator scripts, written content, or generated visual assets; linked third-party services and badges remain subject to their respective terms.
| Evidence-backed measure | Current repository evidence |
|---|---|
| Generated profile assets | The asset builder deterministically renders 10 SVG files from versioned profile content. |
| Automated checks | 9 tests cover profile data, SVG helpers, deterministic generation, bounded telemetry fallback behaviour, and HTTPS-only remote assets; CI also verifies Ruff and a clean rebuild. |
The qualitative outcome is a reproducible, accessibility-aware GitHub profile rather than a hand-edited collection of graphics. These figures describe repository automation, not audience reach or hiring impact.
constfishman7337={name: "Goh Kun Ming",handle: "fishman7337",base: "Singapore",timezone: "Asia/Singapore (SGT, UTC+08:00)",education: "Diploma in Applied AI & Analytics, Singapore Polytechnic",currentRole: "Applied AI Research Intern",publicResearch: "Quantum-Enhanced GANs preprint · arXiv:2508.09209",operatingMode: "research → reproducibility → evaluation → deployable systems",proofPoints: ["PyTorch","Keras","Qiskit","Pandas/NumPy","SQL","AWS","Docker","CI/CD"]};I’m focused on a practical arc: ask a research question, build a reproducible baseline, measure it honestly, then turn the useful pieces into documented systems.
The common thread across the work below is quantum ML, computer vision, geospatial preparation, MLOps, and responsible release discipline.
Hybrid quantum-classical ML, GAN baselines, reproducible experiments, and evaluation discipline. | Computer vision, satellite imagery preparation, sensor fusion, remote-sensing features, and detection pipelines. | Flask/FastAPI apps, Docker workflows, CI/CD, model serving, tests, security checks, and MLOps docs. | Responsible AI thinking, model cards, data cards, threat models, and accountable release practices. |
🧩 Expand the operating principles
| Principle | How I apply it |
|---|---|
| Truth-grounded claims | I prefer honest baselines, clear limitations, and reproducible evidence over inflated results. |
| Systems thinking | A model is only useful when the surrounding data, testing, deployment, monitoring, and docs are coherent. |
| Research-to-product loop | I like converting experiments into usable, reviewable, well-documented artifacts. |
| Safety and governance | I treat documentation, access control, threat modeling, and risk controls as first-class engineering work. |
Comparative Analysis of Classical and Hybrid Quantum-Classical Generative Adversarial Networks My arXiv preprint investigates whether parameterised quantum circuits can act as useful latent priors for generative modelling under near-term quantum constraints. The work compares a classical GAN against multiple HQCGAN variants and keeps the claims bounded by image-quality metrics and limitations.
|      Quantum AI × Generative Modelling × Evaluation Discipline |
🔬 Open the research mental model
HQCGAN experiments comparing classical GAN baselines with noisy quantum-circuit latent priors on binary MNIST. Signal stack: Qiskit · TensorFlow · GANs · FID/KID · experiment configs · reproducibility · arXiv. | Satellite imagery preparation, geometry/topology feature engineering, preliminary model screening, and W&B-tracked orchestration. Signal stack: remote sensing · computer vision · PyTorch · W&B · geospatial features · governance docs. |
Historical security analytics, governance/public-policy panels, ML/DL, graph intelligence, RAG safety, and reproducible MLOps. Signal stack: React · ML · graph analytics · Neo4j · data engineering · RAG guardrails. | Vegetable image classifier with model serving, auth, prediction history, CI/CD, pytest, security checks, Docker, and MLOps docs. Signal stack: Flask · TensorFlow serving · model registry · CI/CD · security scanning · operational docs. |
Open the full project map
| Arena | Repository | What it demonstrates |
|---|---|---|
| 🧬 Quantum / Generative AI | hybrid-quantum-classical-gan-research | HQCGAN research, noisy quantum circuits, GAN evaluation, reproducible experiments |
| 🛰️ Geospatial / ISR | ISR | Satellite imagery preparation, feature engineering, PyTorch screening, W&B orchestration |
| 🌐 Policy Intelligence | global-security-policy-intelligence | Historical analytics, graph intelligence, RAG safety, governance panels |
| 🥬 MLOps Product | sp-daaa-doaa-ca2-vegetable-classification-application | Model serving, CI/CD, security scans, Docker, classification app |
| 🏙️ Multimodal ML App | sp-daaa-doaa-ca1-housing-price-ml-application | Tabular + NLP + image signals, Flask, Docker, tests, MLOps docs |
| 🔐 Secure Systems | yubikey-secure-endpoint-system | Rust endpoint watchdog, security-key checks, audit logging, threat thinking |
| 📈 Math + Regression | sp-daaa-mai-ca3-wage-modelling | Regression modelling, gradient descent, pytest, LaTeX reporting |
| 💬 NLP / Deep Learning | sp-daaa-dele-ca1-movie-review-sentiment-analysis | RNN, LSTM, GRU sentiment/rating prediction workflows |
| 🕹️ Reinforcement Learning | sp-daaa-dele-ca2-pendulum-reinforcement-learning | DQN-style experimentation and control-task learning |
| 📊 Visual Analytics | sp-daaa-davi-ca1-hdb-price-dashboard | HDB resale analytics, Tableau workbook, cleaning/validation scripts |
Clickable stack map - rounded logo cards - no duplicate technologies
Open the capability matrix
| Capability | Tools / methods | Portfolio signal |
|---|---|---|
| Machine Learning | Python, PyTorch, TensorFlow, Keras, scikit-learn, Jupyter, evaluation metrics | Classification, regression, GANs, RL, practical ML apps |
| Web + Frontend | HTML5, CSS, JavaScript, TypeScript, React, Vue, Next.js, Node.js, Vite, Tailwind CSS | Dashboards, portfolio UI, public-facing AI tools, interactive app surfaces |
| Data + Statistics | Pandas, NumPy, Matplotlib, Seaborn, statsmodels, SQL | HDB analytics, wage modelling, statistical reports, validation scripts |
| Vision + Robotics | OpenCV, ROS, sensor fusion, satellite imagery, feature engineering | ISR, CV classification, object detection, perception pipelines |
| Quantum AI | Qiskit, PennyLane, AerSimulator, parameterised quantum circuits, NISQ-aware design | HQCGAN research and arXiv preprint |
| Visual Apps + Graphs | Plotly, Dash, Tableau, Power BI, Neo4j, graph analytics | Dashboards, policy intelligence, exploratory analysis, graph-backed insights |
| Systems + Cloud | Flask, FastAPI, Docker, Kubernetes, GitHub Actions, AWS, GCP, pytest | Model serving, CI/CD, security checks, deployable apps |
| Governance + Security | threat models, audit logs, model cards, risk controls, Rust | Responsible AI docs and secure endpoint tooling |
Telemetry timezone: Singapore Time (SGT, UTC+08:00).
Cards are checked-in SVG snapshots for reliable GitHub rendering. Refresh with python scripts/build_github_telemetry_cards.py.
