(Senior) Machine Learning Scientist

About us Gogolook is a leading TrustTech company founded in 2012 and listed on the Taiwan Stock Exchange in 2025 under the stock code 6902. With "Build for Trust" as its core value, the company has expanded its business from Asia to Europe and America. Gogolook’s AI technology is built on the world's largest database of digital scam data, encompassing phone numbers, websites, virtual currency wallet addresses, and other factors. The company provides diverse anti-scam and fintech services for both consumers and businesses. Its anti-scam offerings include the digital anti-scam app "Whoscall" and a range of enterprise scam prevention solutions in combination with "ScamAdviser." The Fintech BU empowers consumers through data-driven financial services and inclusive lending: Roo.Cash（袋鼠金融）offers transparent matchmaking for financial products like credit cards, while JUJI（招財麻吉）provides an innovative microloan service for rapid, convenient funding during urgent financial needs. A foundation member of the Global Anti-Scam Alliance (GASA), Gogolook has also teamed up with a number of institutes such as the Taiwan National Police Agency Criminal Investigation Bureau, the Financial Supervisory Service of South Korea, Thai Royal Police, the Fukuoka city and Shibuya City government, the Philippines Cybercrime Investigation and Coordinating Center, and the Royal Malaysia Police and state government to fight scam, dedicated to creating a "scam-free environment." Why you should join Gogolook Influential products: What we make are meaningful products that create values for society and defend against frauds. Emphasize self-growth: We encourage technical community activities, subsidize tickets for conferences and workshops so that learning is continuously supported by the company. Unleash your talent: We respect the professional opinions of everyone, encourage team members to discuss with each other, and make awesome products together. Transparent culture: We publicly share the company's information to all, every member can read and feedback, and become a part of participating in the proposal.

Responsibilities

• Hypothesis-Driven Modeling: Frame product questions as testable hypotheses; choose and develop models (supervised, unsupervised, time-series, probabilistic) on mathematical grounds and evidence, not convention or intuition.
• Experimental Design: Design, run and interpret experiments — hypothesis tests, confidence intervals, A/B tests, multi-armed bandit, factorial and stratified designs — to decide what to change and when to stop.
• Rigorous Evaluation & Error Analysis: Choose metrics that reflect real cost (e.g. precision–recall trade-offs under class imbalance), quantify uncertainty, and run error analysis that explains where and why a model fails.
• Validity & Robustness: Detect and guard against overfitting, data leakage, selection bias and model drift; establish how a model will behave as scam patterns shift.
• Mathematical & Empirical Justification: Defend every significant decision from first principles and data — the question, the mathematical reasoning, the design, the assumptions, the evidence and the conclusion — written up so that others can reproduce and challenge it.
• Path to Production: Hand validated models to engineering teams with clear acceptance criteria, and define the monitoring that shows whether the model still holds in production.
• Scientific Leadership: Review peers' experimental designs, mentor on statistical methods, shape the research agenda, and define team-wide standards for reproducibility and rigor.

Qualifications

• Proficiency in English, both spoken and written, sufficient to collaborate, document, and present in an English-working environment.
• Graduate degree (Master's or PhD) in AI, Machine Learning, Deep Learning, or a related or supporting field such as Statistics, Mathematics, Physics or Computer Science with an ML focus.
• 6–10 years of experience in machine learning research or applied ML.
• Strong mathematical foundations: probability and statistics, linear algebra, and optimization, applied fluently to real modeling problems.
• A scientific approach to problems: you form hypotheses, test them, and change course when the evidence says so — and can show examples from your past work.
• Mathematical defense of choices: you justify a model, loss, metric or experimental design by its mathematical properties and assumptions, not because it sounds logical or is what others use.
• Evidence-driven communication: you can explain and justify a modeling direction to both technical and non-technical audiences using mathematics and data, not intuition or authority.
• Independent and self-motivated, able to own an investigation from open question to defensible conclusion.
• Proactive communicator who raises methodological and technical risks early.

Preferred Qualifications

• PhD, or a record of peer-reviewed publications or rigorous technical reports in ML or statistics.
• Experience with Bayesian methods, causal inference, or anomaly detection.
• Experience with adversarial or non-stationary domains such as fraud, abuse, spam or security.
• Exposure to cloud ML platforms (AWS SageMaker, Google Vertex AI), and enough familiarity with CI/CD and Docker to hand models over to production teams smoothly.

Skills & Competencies

• Python and the scientific stack (NumPy, pandas, SciPy, statsmodels); SQL for working directly with large datasets.
• ML frameworks: scikit-learn, XGBoost, and PyTorch or TensorFlow.
• Statistical toolkit: hypothesis testing, confidence intervals, power analysis, A/B and multi-armed bandit testing, and resampling methods.
• Git and reproducible experiment tracking.

How to Apply

If you have published articles or papers on ML or AI, please share them with your application. We will ask you to walk us through them during the interview: the question, the method, and how you justified your conclusions.