Recommendation Algorithm Engineer

Stealth Web3 Startup; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

Stealth Web3 Startup seeks an experienced ML Engineer to build end-to-end personalized recommendation systems for prediction markets, covering home feeds, trending events, and market data insights. You will craft user profiles, design the full pipeline from retrieval to ranking, and optimize for engagement, retention, and trading actions.

Importantly, you will drive data-driven iterations, evaluate experiments, and implement robust monitoring to balance accuracy with content diversity and user

Qualifications

• 5+ years of hands-on experience in production search or recommendation systems.
• Experience with content feeds, news, trending events, financial market data, or community recommendation systems is preferred.
• Strong understanding of the complete recommendation system architecture, including retrieval, ranking, re-ranking, multi-objective optimization, and cold-start strategies.
• Proficient in PyTorch or TensorFlow, with ability to independently conduct feature engineering, model training, offline evaluation, online tuning, and iterative experimentation.
• Experience with user behavior sequence modeling and handling sparse data.

Responsibilities

• Build and optimize the end-to-end personalized recommendation system for a prediction market platform.
• Develop multidimensional user profiles based on browsing, clicks, follows, trades, dwell time, and search behaviors to model user interests and activity levels.
• Design and optimize the full recommendation pipeline, including cold start, candidate retrieval, ranking, diversity, and real-time relevance.
• Build a scalable feature engineering framework from mining user behavior sequences.
• Continuously improve recommendation models through data-driven iterations and establish robust evaluation and monitoring mechanisms.

Skills

Recommendation systems

Deep learning

A/B testing

Vector retrieval

User profiling

Education

Bachelor's degree or above

Tools

PyTorch

TensorFlow

Job description

• Build and optimize the end-to-end personalized recommendation system for a prediction market platform, covering personalized home feeds, trending events, sports/event markets, and other recommendation scenarios. Drive improvements in key business metrics, including CTR, user engagement time, participation rate, retention, and trading conversion.
• Develop multidimensional user profiles based on browsing, clicks, follows, trades, dwell time, and search behaviors to model user interests, risk preferences, thematic preferences, and activity levels.
• Design and optimize the full recommendation pipeline, including cold start, candidate retrieval, ranking (CTR/CVR prediction), diversity, multi-objective optimization, real-time relevance, and trending content boosting.
• Build a scalable feature engineering framework by mining user behavior sequences and constructing high-quality recommendation features.
• Continuously improve recommendation models and strategies through data-driven iterations, balancing personalization accuracy, content diversity, popularity concentration, and overall user experience. Establish robust evaluation, monitoring, and optimization mechanisms for recommendation performance.

Qualifications

• Bachelor's degree or above in Computer Science, Artificial Intelligence, Data Science, or a related field, with 5+ years of hands-on experience in production search or recommendation systems. Experience in content feeds, news, trending events, financial market data, or community recommendation systems is highly preferred.
• Strong understanding of the complete recommendation system architecture, including candidate retrieval (vector-based, rule-based, and popularity-based), ranking, re-ranking, multi-objective optimization, and cold-start strategies. Proficient in mainstream CTR/CVR prediction techniques.
• Proficient in deep learning frameworks such as PyTorch or TensorFlow, with the ability to independently conduct feature engineering, model training, offline evaluation, online tuning, and iterative experimentation.
• Solid experience in feature engineering, user behavior sequence modeling, and techniques for handling sparse data.
• Strong analytical skills and business understanding, capable of identifying recommendation issues through data analysis (e.g., traffic imbalance, insufficient long-tail exposure, ineffective cold start, content homogenization) and implementing effective optimization strategies.
• Familiar with A/B testing methodologies and experimental design, with the ability to quantitatively evaluate recommendation strategies and drive continuous optimization.

Preferred Qualifications

• Experience in recommendation systems for financial markets, trading platforms, market data, trending events, news feeds, or sports content is highly preferred.
• Experience with vector retrieval (ANN/Embedding Search), personalized ranking lists, real-time trending ranking, and multi-objective recommendation modeling.
• Experience designing user profiling and tagging systems, with the ability to develop recommendation strategies that combine long-term and short-term user interests.