Applied AI Engineer

Hasbro•Boston, MA

21h•Onsite

About The Position

We are looking for curious adventurers ready to find their party, fueled by imagination and drive to build what’s never been built before. At Hasbro and Wizards of the Coast, you’ll collaborate with passionate teams to reimagine our iconic brands and create experiences that spark joy, connection, and community through the magic of play. This is your chance to shape legendary play that lasts a lifetime. We're building something new inside our AI Studio, and we're looking for an Applied AI Engineer to help shape it with us. This role sits at the center of a new and constantly evolving set of challenges. The problems are still being defined, and the approaches are actively being invented. What excites us is the opportunity to determine what it truly means for an iconic character to think, respond, and stay true to itself - consistently, safely, and at scale. We work where storytelling, imagination, and technology converge. Our focus is on the intelligence layer: the systems that determine how characters understand context, form responses, maintain a consistent voice, and hold carefully defined boundaries. This is where canon knowledge meets language model behavior, and where the quality of every character interaction gets defined and measured. This role owns one of the most creatively and technically interesting areas on the team: making characters that actually feel like themselves. Working closely with engineers, creative partners, and brand leaders, the Applied AI Engineer will define prompt architecture, own guardrails and evaluation frameworks, and build the retrieval systems that ground characters in their own lore. Success here means developing the tools and practices that make character fidelity measurable.

Requirements

3+ years of experience building and shipping production LLM-powered systems.
Deep expertise in prompt engineering — system prompt design, few-shot techniques, chain-of-thought, multi-turn conversation management, and structured output.
Experience designing and implementing evaluation frameworks for LLM outputs, including adversarial and security testing.
Hands-on experience with RAG architectures: embedding models, vector stores, chunking strategies, and retrieval evaluation.
Proficiency with major LLM APIs (OpenAI, Google Gemini, Anthropic, or equivalent) and experience making principled decisions between them.
Strong Python skills and comfort working in AWS environments.
Familiarity with how voice AI services work (TTS, STT) — enough to understand how spoken delivery shapes prompt and response design.
The ability to communicate AI system behavior clearly to creative, brand, and non-technical collaborators.

Nice To Haves

Familiarity with guardrails frameworks (NVIDIA NeMo Guardrails, Guardrails AI, or equivalent).
Conceptual understanding of fine-tuning and RLHF, even if not a primary area of focus.
Background in entertainment, gaming, or IP-driven product development.

Responsibilities

Design, implement, and continuously refine system prompts, character behavior definitions, and guardrails for our AI character platform.
Build evaluation frameworks to measure character fidelity, response quality, and safety — establishing quantitative standards for what "in character" actually means across interaction types.
Architect and implement RAG systems to ground character responses in canonical knowledge, including embedding strategies, vector retrieval, and relevance evaluation.
Partner with creative and brand teams to translate character attributes — personality, voice, canon, tone, and boundaries — into precise, testable AI behavior and system requirements.
Lead prompt versioning, regression testing, and model evaluation practices to ensure character behavior remains consistent across provider updates and model changes.
Integrate with multi-provider LLM APIs (OpenAI, Google Gemini, etc.) and make principled selection decisions as the landscape evolves.
Evaluate and adopt emerging techniques in prompt engineering, agentic design, and LLM evaluation that meaningfully advance the quality and reliability of character experiences.