AI Model Trainer - Business / Technology

FROST & SULLIVAN (SINGAPORE) PTE LTD; Singapore; On-site; SGD 90,000 - 120,000; Full time

Job summary

MetaBrain, a solution developed by Frost & Sullivan, seeks experts to turn industry knowledge into reusable software-enabled advisory services. You will curate research assets and standards, evaluate outputs for accuracy, and continuously improve end-to-end workflows, collaborating with technical trainers and data specialists.

The role emphasizes hands-on research in AI-enabled analytics, rigorous quality assurance, and iterative testing of advisory materials with clear documentation and sources.

Qualifications

• Substantial hands-on research or knowledge-quality experience (typically 3+ years).
• Strong evidence assessment and business writing skills.
• Basic understanding of AI/LLMs from formal learning or a certificate.
• For the technology variant, Python evaluation scripting is required.

Responsibilities

• Build domain learning assets: curate research, taxonomies, decision rules and examples with sources.
• Evaluate advisory quality: write rubrics and document errors and adjudication.
• Improve the workflow: configure prompts, retrieval settings and change requests.
• Connect business and technology: partner with trainers and, for the tech variant, build Python evaluation scripts.

Skills

Research experience

Analytical writing

AI/ML basics

Education

Certificate in AI/ML

Tools

SQL

Python

Job description

Bring your research and consulting expertise into MetaBrain solution development. Turn industry knowledge and advisory methods into reusable, continuously updated software-enabled services.

MetaBrain combines Frost & Sullivan’s 60 years of industry expertise, enterprise data, and third-party insights, delivering human-reinforced AI insights and analytics to solve your toughest business use cases with actionable, real-time intelligence.

https://metafrost.ai/

Key responsibilities

• Build domain learning assets. Curate authorized research, taxonomies, decision rules, examples and counterexamples. Create reference answers with supporting sources and record scope, dates, limitations and permitted use.
• Evaluate advisory quality. Write rubrics for factual support, numerical consistency, relevance, completeness and appropriate uncertainty. Run blind comparisons and document errors, reviewer disagreements and adjudication; keep held-out evaluation cases separate.
• Improve the workflow. Configure prompts, retrieval settings and business rules in approved Design/Tuning environments. Turn advisor corrections into versioned change requests; retest after each update and retain a clear rollback record.
• Connect business and technology. Partner with technical trainers on experiment design and failure diagnosis. In the technology variant, additionally build Python evaluation scripts, manage dataset versions and support approved fine-tuning experiments where justified.

Essential requirements

• Substantial hands‑on research, consulting or knowledge‑quality experience; typically 3+ years, with equivalent achievement considered.
• Strong evidence assessment, business writing, numerical checking and attention to methodological detail.
• Basic understanding of AI/large language models from formal learning or a certificate and a hands‑on example of testing or improving AI outputs.
• Preferred: Experience in peer review, analyst coaching, survey coding, tax onomy creation, quality assurance, low‑code automation or maintaining a structured knowledge base. SQL or Python is helpful for business‑track applicants; Python evaluation scripting is required for the technology variant.
• Entry level and authority: This is not only data labelling or prompt writing. Business feedback may improve sources, rules, retrieval or configuration; it does not automatically retrain a foundation model. Model-weight changes require approved data rights, engineering ownership and regression testing.