AI Anime Researcher - LLM

Job Description

Job Description

tl;dr: We're building a 3D first-person adventure game where an AI companion is the core of the game. Think MiSide with LLMs, but truly integrated with gameplay, not just a RP chatbot.

Here at Spellbrush , we're passionate about making a good anime game.

We also happen to be the world's leading generative AI studio—we're the team behind niji・journey .

Our mission is simple: use AI to bring characters to life and push the boundaries of narrative-driven games.

We’ve developed an in-house LLM storytelling system that blends AI, story, and gameplay—going far beyond shallow “chat-only” experiences.

The result is an AI companion who cooperates with you in puzzles, remembers across worlds, and changes how each chapter unfolds.

In-game screenshots (development build)

You will join a small but extremely talented team to define a new kind of video game experience. You'll work alongside top otaku minds in the industry, including the creator of Warudo and Cytoid, a Google Deepmind veteran behind Project Astra, and top-tier AI researchers.

As an early member of this team, you will have significant artistic and research freedom to shape what could become the next-generation LLM-driven storytelling experience.

Impact and Ownership: As an early team member, you'll have the opportunity to push the boundaries of what's possible at the intersection of the AI, anime, and video game industry.

A World-Class Team: You'll work directly with some of the best AI researchers and engineers in the world in a small, mighty, and fast-paced team.

Unparalleled Resources: You will have a generous amount of compute to build the best possible models for anime characters.

Culture: We are seeking individuals who ideally are fans of anime and the anime aesthetic. We believe in the unmatched speed of in-person collaboration.

International Opportunities: We have primary research offices in downtown Akihabara, Tokyo, and San Francisco. Visa sponsorships are available for qualified candidates

You have a deep understanding of LLM training and evaluation with attention to detail.

Bonus if your post training work is focused on roleplay.

Bonus if you have worked on latency optimization. We need to interact with the character in real time.

Whether that is by designing a RLHF-style voting system, or by finding implicit signals in the human-AI interaction logs.

Training LLMs requires both research and engineering. Experiences in building data processing backend, deterministic and checkpointable data loader, sharding, memory saving techniques, scalable benchmarking and checkpoint selection, are all helpful.

As a member of our team, you'll have the opportunity to push the boundaries of what's possible in the anime and video game industry. Ideally you're a fan of the genre.

Our engineering team is small but mighty. You'll be working directly with some of the best AI researchers and engineers in the world.

We also believe in the unmatched speed of in-person teams, and prefer on-site collaboration in either our primary research office in Tokyo (downtown Akihabara) or San Francisco. Visa sponsorships are available.

Or, putting it in RL jargon, exploration and exploitation. You excel at working with research teams to synthesize high-impact needs, design and implement technical solutions, and communicate deliverables and tradeoffs.