Search & Recommendation Algorithm Engineer

JonDavidson Pte Ltd; Singapore; Sur place; SGD 180 000 - 240 000; Plein temps

Résumé du poste

JonDavidson Pte Ltd seeks an experienced Search & Recommendation Algorithm Engineer to drive end-to-end R&D and system architecture for an overseas AIOS platform. You will design high-precision, multilingual search and recommendation systems for global household users, leveraging large-scale online systems and advanced ML models.

The role requires deep expertise in deep learning models, vector databases, and end-to-end data loops with A/B experimentation to validate strategies and iterate

Qualifications

• 5+ years of R&D experience in Search, Recommendation, or NLP algorithms deployed in large-scale online systems.
• Strong foundation in deep learning models such as Two-Tower Recall, DeepFM, Wide&Deep, DIN, and Multi-Objective Ranking.
• Proficient in Python and at least one framework (PyTorch or TensorFlow).
• Experience with search engines and vector databases (Elasticsearch, Faiss, Milvus).
• Solid experience in feature engineering, user behavior sequence modeling, and real-time/offline computing frameworks (Flink / Spark).
• Familiarity with evaluation metrics (CTR, CVR, Recall, NDCG) and A/B experimentation methodologies.

Responsabilités

• Full-Link Search Development: design and iterate search algorithms including intent understanding, multi-lingual query parsing, multi-path recall, and ranking.
• Personalized Recommendation Systems: build dual-layer user profiling and recommendation models across recall, coarse ranking, fine ranking, and re-ranking stages.
• Unified Architecture Integration: lead architecture evolution uniting real-time recommendation and search.
• Global Multi-lingual & Vector Search: optimize cross-lingual semantic retrieval and unified vector embeddings for voice and text search.
• Data Closed-Loop & Experiments: build end-to-end data loops and run continuous A/B testing to validate performance.

Connaissances

Full-Link Search

Multi-lingual query processing

Vector embeddings

A/B testing

Data pipelines

Formation

Bachelor's degree or higher (Master’s preferred)

Outils

Python

PyTorch

TensorFlow

Elasticsearch

Faiss

Milvus

Flink

Spark

Description du poste

We are seeking a Search & Recommendation Algorithm Engineer to drive end-to-end algorithm R&D and system architecture for our overseas AIOS platform. You will build high-precision, multi-lingual, and family-scenario-aware search and recommendation systems targeting global household users. This role requires deep expertise in large-scale online systems, machine learning models, and joint search-recommendation modeling.

Key Responsibilities

• Full-Link Search Development: Design and iterate full-link search algorithms including intent understanding, multi-lingual query parsing, multi-path recall, fine ranking, and re-ranking to optimize CTR, conversion, and satisfaction metrics.

• Personalized Recommendation Systems: Build a dual-layer user profiling system (Household Group + Individual Member). Develop recommendation models across recall, coarse ranking, fine ranking, and re-ranking stages to solve multi-user interest conflicts and cold-start challenges in home viewing scenarios.

• Unified Architecture Integration: Lead the architecture evolution uniting real-time recommendation and search. Standardize content representations, user modeling, and ranking systems to enable bidirectional data flow between search intent and recommendations.

• Global Multi-lingual & Vector Search: Optimize cross-lingual semantic retrieval, multi-lingual query processing, and unified vector embeddings for voice and text search across global markets.

• Data Closed-Loop & Experiments: Build end-to-end data loops (exposure-click-conversion-feedback) and run continuous A/B testing to validate strategy performance and iterate algorithms.

Key Requirements

• Education & Experience: Bachelor’s degree or higher (Master’s preferred) in Computer Science, Math, Software Engineering, or related fields. 5+ years of R&D experience in Search, Recommendation, or NLP algorithms with proven deployment in large-scale online systems.

• Machine Learning Models: Strong foundation in deep learning models such as Two-Tower Recall, DeepFM, Wide&Deep, DIN, and Multi-Objective Ranking.

• Tech Stack & Frameworks: Proficient in Python and at least one framework (PyTorch or TensorFlow). Practical experience with search engines and vector databases (Elasticsearch, Faiss, Milvus).

• Data & Compute Pipelines: Solid experience in feature engineering, user behavior sequence modeling, and real-time/offline computing frameworks (Flink / Spark).

• Business Acumen: Familiarity with evaluation metrics (CTR, CVR, Recall, NDCG) and statistical A/B experimentation methodologies.

Preferred Qualifications

• Prior search/recommendation algorithm experience on TV platforms, smart home terminals, or streaming media services.
• Expertise in multi-lingual NLP, cross-lingual retrieval, LLM retrieval augmentation (RAG), or household multi-user modeling