NLP Research Engineer for Financial Disclosures & RAG

Singapore Institute of Technology; Singapore; On-site; SGD 56,000 - 89,000; Full time

Job summary

Singapore Institute of Technology invites a Research Engineer to join a MOE AcRF Tier 1 funded project spanning accounting and NLP/AI. You will design, build, and validate pipelines that detect inconsistencies between financial filings and earnings call transcripts, using retrieval-augmented NLI models and LLMs on large-scale financial text data.

The role involves implementing retrieval systems, processing data, and supporting empirical analysis of contradiction scores, with emphasis on

Qualifications

• Bachelor’s or Master’s in CS/IS/AI or related field.
• Strong Python programming and NLP foundations.
• Familiarity with ML/NLP libraries and model deployment concepts.
• Experience with version control (Git) and collaborative projects.

Responsibilities

• Drive the technical execution of the research project under PI/Co-PI guidance.
• Design and implement retrieval-augmented NLI pipeline for detecting inconsistencies across disclosures and transcripts.
• Preprocess large-scale financial text data and index corpora.
• Implement and evaluate claims extraction using LLM-based approaches.
• Develop semantic retrieval for claim-context matching across channels.
• Build and assess contradiction detection models using NLI/LiM prompting.
• Assemble firm-quarter level scores and datasets for analysis.
• Coordinate with student annotators and prepare annotation guidelines.
• Maintain reproducible workflows with version control and documentation.

Skills

Python programming

NLP

Machine learning

Information retrieval

Git

Team collaboration

Education

Bachelor’s or Master’s in Computer Science or related field

Tools

HuggingFace Transformers

PyTorch

TensorFlow

scikit-learn

Job description

Singapore Institute of Technology invites a Research Engineer to join a MOE AcRF Tier 1 funded project spanning accounting and NLP/AI. You will design, build, and validate pipelines that detect inconsistencies between financial filings and earnings call transcripts, using retrieval-augmented NLI models and LLMs on large-scale financial text data.

The role involves implementing retrieval systems, processing data, and supporting empirical analysis of contradiction scores, with emphasis on