Applied AI Engineer Intern - Advancing Agent Quality

Role Type: External Research Affiliate / Think Tank Contributor
Location: South Florida preferred
Affiliation: Local university-based Masters or PhD researcher or postdoctoral AI researcher working under faculty direction
Organization: Brightstar.

AI

Brightstar.

AI is seeking a full-time or recently graduated Masters or PhD-level Applied AI Engineer Intern –
Advancing Agent Quality with a strong research mindset, affiliated with a leading (South Florida) university, to contribute to its AI Think Tank and to the hands-on build-out of Brightstar’s proprietary AI systems. The intern will pair academic rigor with applied engineering: building the evaluation environments, automated raters, and diagnostic tooling that establish – with evidence rather than impression – whether an agent is reliable enough to put in front of operators, executives, and investors.

The first concrete assignment is the AI Twin Board
, Brightstar’s proprietary board-simulation product, where the intern will help build the evaluation harness end to end: realistic task suites, calibrated LLM judge, failure-mode diagnostics, and the data flywheel that feeds improvements back into the agents themselves.

Working in coordination with university professors and research leaders, the intern will also bring up-to-date knowledge of the fast-evolving AI landscape — especially large language models, reasoning systems, emerging model capabilities, and the societal implications of advanced AI — and help translate what current and next-generation systems may enable over the next six months to three years into Brightstar’s evaluation standards and Think Tank point of view.

• Design, build, and scale realistic agent environments and task suites that reflect how Brightstar’s agents are actually used — board-level deliberation, research synthesis, diligence workflows, and multi-step tool use.
• Develop agentic LLM judge, calibrate them against human expert evaluation, and benchmark Brightstar agents against leading frontier models.
• Build trajectory-analysis frameworks and diagnostic tooling that root-cause agent failure modes — passivity, hallucination, persona drift, brittle tool execution — and feed fixes back into agent prompts, harnesses, and architecture.
• Enable agents to generate and iterate on their own verifiers (automated test cases, checklists, ground-truth sets) to support effective exploration and iterative problem-solving.
• Harvest multi-turn interaction trajectories into high-quality datasets and reward signals that support fine-tuning and post-training of the models Brightstar relies on.
• Contribute to the AI Twin Board build-out as the initial focus, and extend the same evaluation approach to other Brightstar AI initiatives as the roadmap grows.
• Contribute frontier AI research perspectives to the Brightstar.
  
  AI Think Tank as part of weekly review sessions and select strategic discussions.
• Review relevant research papers, conference findings, and publications from leading AI institutions and labs as inputs to the Think Tank knowledge base and to Brightstar’s evaluation methodology.
• Help interpret academic and technical developments for a business and investment audience, including implications for industry transformation and applied AI opportunities.

• Currently enrolled in a Masters or PhD program — or serving as a postdoctoral or early-career academic researcher — in Artificial Intelligence, Machine Learning, Computer Science, Computational Linguistics, or a related field
  .
• 1 years of research, project, or professional experience with LLMs or Agents.
• 2 year of experience with machine learning and deep learning.
• Experience in software engineering and cloud-based development.
• Experience with…