Jerry Tay

Data Science & AI Graduate with hands-on experience building production-ready Conversational AI, Retrieval-Augmented Generation (RAG) systems, and Knowledge Graphs. Proven track record in evaluating LLM performance via RAGAS, engineering hybrid vector-graph retrieval pipelines, and deploying containerized microservices via FastAPI and Docker. Co-author of a published Springer research paper on multimodal misinformation detection.

AI / Data Science Intern at Sarawak Information Systems Sdn. Bhd. (SAINS) (2025-06 – 2025-12)

• LLM & RAG System Engineering: Developed AI-powered chatbots on the Dify platform using custom RAG pipelines, contextual chunking strategies, and metadata enrichment to minimize model hallucination.
• Hybrid GraphRAG Implementation: Engineered a hybrid retrieval prototype integrating Neo4j knowledge graphs, semantic vector search, and keyword retrieval using FastAPI and Docker to enable multi-hop entity reasoning.
• Automated Evaluation Pipelines: Built end-to-end evaluation workflows using Python and RAGAS, leveraging synthetic persona-based QA datasets to assess Context Recall, Faithfulness, Answer Relevance, Response Correctness, and Context Precision.
• Quality Benchmarking: Enforced a strict minimum quality threshold of 0.70 (70%) across all RAGAS evaluation metrics for deployed chatbot agents.
• API Integration & Testing: Built and exposed microservices for real-time news retrieval, multi-bill queries, and external knowledge graph expansion.
• Cross-Functional Collaboration: Participated in daily Agile stand-ups, sprint reviews, and communicated technical ML architectures to non-technical stakeholders.

Bachelor (Hons) in Data Science – Tunku Abdul Rahman University of Management and Technology (TAR UMT) (2022 – 2025-12)