Senior AI Lead, Multimodal Systems

Track: Senior IC with a clear path to Chief AI Officer (CAIO) AI domains: LLM · Image Gen · Video Gen · Sound & TTS · Agents · Model Training · Fine-Tuning · Infrastructure · Evaluation · Data Pipelines Languages: Japanese required (business level); business-level English or Mandarin also required

About the role

We are hiring a Senior AI Lead — a technically exceptional generalist who can architect and orchestrate the full spectrum of AI modalities needed to bring a living, breathing companion game to life. This is not a narrow specialist role: you will own the strategy and execution across language, vision, audio, motion, and agent behavior, making them work as a seamless, immersive whole.

This is one of the most senior and consequential roles at Gazai. You will define the AI architecture of Anini, lead a growing AI team, and sit on the technology leadership team with a clear path toward Chief AI Officer. You will shape not just what we build, but how we think about AI at Gazai.

Responsibilities

• Architect and lead the multimodal AI system powering Anini — integrating LLM dialogue, image generation, video synthesis, sound and voice, and autonomous agents into a cohesive, real-time companion experience
• Define and own the long-term AI roadmap across all modalities; translate product vision into concrete AI research and engineering priorities
• Lead model selection, fine-tuning, and post-training across domains — including character-consistent image generation, expressive TTS, story-aware LLMs, and behavioral agents
• Design and oversee agent architectures enabling proactive, autonomous companion behavior: planning, memory, tool use, and real-world integrations (social media, smart home, IoT)
• Establish evaluation frameworks and quality standards across all AI outputs — latency, coherence, visual consistency, emotional expressiveness, and safety
• Build and manage scalable AI infrastructure: model serving, data pipelines, training compute, and cost optimization
• Grow and mentor the AI team; set engineering culture and best practices across the function
• Collaborate across product, engineering, and leadership to deliver AI innovations to customers
• Track the frontier of AI research across all relevant modalities and rapidly prototype what matters

Qualifications

• 7+ years of ML/AI engineering experience, including leadership of AI systems or teams
• Hands-on depth in at least two AI modalities (e.g. LLMs + image gen, or agents + video synthesis)
• Strong conceptual and practical understanding of modern deep learning — transformers, diffusion models, autoregressive generation
• Experience fine-tuning or post-training large models (RLHF, DPO, LoRA, etc.)
• Experience designing and shipping agentic systems using frameworks such as LangGraph, AutoGen, CrewAI, or custom-built architectures
• Proficiency in Python; comfort with model serving infrastructure (e.g. vLLM, Triton, Ray Serve)
• Strong instinct for system design: latency, reliability, and cost tradeoffs at scale
• Ability to lead cross-functional AI projects and communicate clearly across research, engineering, and product
• Japanese required (business level); business-level proficiency in English or Mandarin also required
• Eagerness to stay at the frontier — fast learner with strong research literacy

Bonus — you will stand out if…

• You have experience with character-consistent image or video generation, style LoRAs, or anime/illustration-specific fine-tuning
• You have shipped real-time or low-latency multimodal pipelines in a consumer product context
• You have experience with voice synthesis, expressive TTS, or sound generation AI
• You have research publications or open-source contributions in generative AI, agents, or multimodal systems
• You have built evaluation infrastructure for generative AI (human evals, automated evals, red-teaming)
• You have experience with vector databases (Pinecone, Qdrant, Chroma) and retrieval-augmented systems
• You have prior experience in the game, entertainment, or interactive media AI space