AI Scientist - Anti-Fraud

DUOTECH PTE. LTD.; Singapore; On-site; SGD 80,000 - 120,000; Full time

Job summary

A leading technology firm in Singapore is seeking experienced AI Scientists to build advanced fraud detection and abuse prevention systems. You will leverage graph machine learning, transformers, and behavioral analytics to tackle identity risk, account takeovers, and more. The ideal candidate has 5+ years of experience in machine learning, is proficient in Python, and is familiar with big data tools. This role requires the ability to communicate in both Chinese and English.

Qualifications

• 5+ years of hands-on experience in machine learning and fraud detection, ideally in fintech, e-commerce, or Web3.
• Able to work with Chinese-speaking stakeholders.

Responsibilities

• Design, build, and deploy AI models using transformer architectures for anomaly detection.
• Build scalable graph learning pipelines to detect malicious actors.
• Create robust behavioral feature extraction layers.
• Implement high-performance fraud scoring systems.
• Collaborate with marketing teams to identify abuse vectors.
• Conduct regular performance evaluations of ML models.
• Support fraud investigation workflows with actionable insights.
• Stay current on the evolving threat landscape.

Skills

Graph machine learning

Transformer models

Fraud detection techniques

Feature engineering

Python

ML frameworks (PyTorch/TensorFlow)

Big data processing tools

Real-time data pipelines

EVM-compatible blockchain data understanding

Bilingual (Chinese and English)

Education

Bachelor’s or Master’s degree in Computer Science, Data Science, or related field

Tools

PyTorch

TensorFlow

Spark

Hive

Kafka

Flink

Job description

We are seeking multiple experienced AI Scientists to build intelligent fraud detection and abuse prevention systems that secure our users and ecosystem. You will leverage the latest advancements in graph ML, transformers, and behavioral analytics to address some of the most critical threats today.

Key Risk Use Cases You'll Tackle

• Identity Risk: Fake KYC profiles, identity farming, synthetic user creation
• Account Takeover (ATO): Compromised credentials, anomalous login behavior, device takeover
• Toll Fraud & Drainers: Gas draining contracts, malicious token airdrops, phishing‑induced approval scams
• Money Laundering: Cross‑chain obfuscation, mixer usage, layering and integration strategies
• Campaign Reward Abuse (Treaty Hunters): Multi‑account farming, airdrop exploitation, referral & cashback fraud across marketing programs

You’ll work closely with engineering, product, compliance, and security teams to build end‑to‑end machine learning systems that detect, explain, and prevent these behaviors in real time.

Responsibilities

• Design, build, and deploy AI models using transformer architectures, GNNs, and unsupervised learning for anomaly and abuse detection.
• Build scalable graph learning pipelines to detect communities of malicious actors and their on‑chain behavioral patterns.
• Create robust behavioral feature extraction layers using on‑chain transaction data, login events, referral logs, device/browser/session metadata.
• Implement high‑performance fraud scoring systems that can operate in real time at massive scale.
• Collaborate with campaign & marketing teams to identify abuse vectors in rewards, referral, and incentive systems.
• Conduct regular performance evaluations of ML models and continuously improve them using new research and adversarial learning techniques.
• Support fraud investigation workflows by providing interpretable model results and actionable insights.
• Stay current on the evolving threat landscape and proactively develop solutions against novel fraud patterns.

Qualifications

Must‑Have

• Bachelor’s or Master’s degree in Computer Science, Data Science, or related field.
• 5+ years of hands‑on experience in machine learning and fraud detection, ideally in fintech, e‑commerce, or Web3.
• Strong background in:
• Graph machine learning, GNNs, and clustering algorithms
• Transformer models and attention mechanisms
• Fraud detection techniques, including anomaly detection and behavioral scoring
• Feature engineering from high‑dimensional, noisy, and heterogeneous data sources
• Proficiency in Python and ML frameworks like PyTorch/TensorFlow.
• Familiarity with big data processing tools (Spark, Hive) and real‑time data pipelines (Kafka, Flink).
• Solid understanding of EVM‑compatible blockchain data (transactions, contract calls, logs) and wallet behaviors.
• Able to use Chinese and English as working language to work with chinese speaking stakeholders

Preferred

• Experience with fraud rule engines or security scoring systems.
• Familiarity with tools like Chainalysis, Tenderly, Alchemy, or similar blockchain analytics platforms.
• Experience modeling campaign abuse behaviors, e.g., multi‑account ring detection, bonus hunting, or engagement spoofing.