Senior Data Engineer — AI & Multimodal Data Pipelines

SINGAPORE TELECOMMUNICATIONS LIMITED; Singapore; On-site; SGD 90,000 - 130,000; Full time

Job summary

Singtel is building a new AI & Data Analytics unit to empower scalable content extraction and intelligent data processing across the enterprise. The role involves designing end-to-end document parsing pipelines, OCR/VLM systems, embedding models, and RAG-enabled knowledge bases on cloud platforms, with emphasis on reliability and governance.

You will collaborate with senior engineers, implement CI/CD, and integrate tools like Kafka, LangChain, PyMuPDF, and Databricks to drive AI-led

Qualifications

• Bachelor’s degree in Computer Science, Engineering, or a related field.
• 1–3 years of experience in data engineering or data platform development.
• Ability to independently build basic batch or streaming document processing pipelines.
• Hands-on experience with Python, PyTorch, PEFT, LoRA, OCR, VLM, LLM, prompting, Text-to-Speech models and SQL for data transformation.
• Familiarity with Apache Spark (PySpark) and large-scale data processing concepts.
• Experience with knowledge base and RAG solutions for agentic AI use cases.
• Self-starter with strong problem-solving skills and collaboration.
• Strong documentation and communication skills.

Responsibilities

• Design, build, productionize, and support scalable document parsing and content extraction pipelines on a modern hybrid cloud platform, handling structured and unstructured data including PDFs, PPTX, Excel, Word, images, scanned documents, videos, audios, charts, tables.
• Design and implement OCR, VLM, and multimodal AI pipelines leveraging state-of-the-art models for accurate content extraction, layout understanding, and table/chart interpretation.
• Develop and optimize embedding models, including fine-tuning and domain adaptation to support organization-specific terminology, keywords, and semantic search use cases.
• Implement knowledge ingestion, indexing, and retrieval pipelines for RAG and agentic GenAI systems, ensuring high recall, precision, and low-latency retrieval.
• Build and maintain data ingestion pipelines for batch and streaming data sources using tools like Kafka, LlamaIndex, LangChain, PyMuPDF, MSAL.
• Implement layout-aware parsing to preserve document structure, hierarchy, and reading order (using OCR, VLM, and multimodal AI solutions).
• Extract accurate numeric and semantic data from complex charts and irregular tables while preventing hallucinations.
• Assist in integrating data from diverse source systems (files, APIs, databases, streaming).
• Implement, fine-tune, and optimize embedding models to support organization-specific terminology, improving semantic search and GenAI performance.
• Help maintain metadata and pipeline documentation for transparency and traceability.
• Participate in integrating pipelines with tools such as Microsoft Fabric, Databricks, Delta Lake, and other platform components.
• Build and maintain knowledge base and RAG solution on cloud (Azure).
• Implement and operate knowledge base storage, lifecycle management and embedding/vectorization.
• Contribute to automation efforts using version control and CI/CD workflows.
• Apply basic data governance and access control policies during implementation.

Skills

Python

PyTorch

PEFT

LoRA

OCR

VLM

LLM

Prompting

SQL

Spark/PySpark

Knowledge base

RAG

Data governance

Documentation

Education

Bachelor’s degree in Computer Science, Engineering, or a related field

Tools

Kafka

LlamaIndex

LangChain

PyMuPDF

MSAL

Microsoft Fabric

Databricks

Delta Lake

Azure

Job description

Singtel is building a new AI & Data Analytics unit to empower scalable content extraction and intelligent data processing across the enterprise. The role involves designing end-to-end document parsing pipelines, OCR/VLM systems, embedding models, and RAG-enabled knowledge bases on cloud platforms, with emphasis on reliability and governance.

You will collaborate with senior engineers, implement CI/CD, and integrate tools like Kafka, LangChain, PyMuPDF, and Databricks to drive AI-led