Technical skills
About this role
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 65 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.
Key responsibilities
Reporting into Lead Architect or AI/ML Lead, with a Frost domain mentor, the incumbent will teach and evaluate how MetaBrain applies approved research methods and business logic. Convert expert knowledge into structured examples, rules and review criteria, and turn advisor feedback into controlled, testable improvements.
- 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, taxonomy 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.