We’re looking for a Senior LLMOps Engineer to join an R&D team of an ambitious deep-tech startup with a ~30-person international team and HQ in Tokyo
The Project
The team is building a natively agentic graph-neuro-symbolic AI system with general capabilities — not just another LLM product, but a foundational AI architecture. The system combines: - principles from mathematical category theory; - knowledge representation; - computational linguistics. The focus is on interpretable, controllable, and safe intelligence, enabling powerful reasoning without turning AI into a black box. This is deep, research-driven work with real production impact.
Responsibilities
Design, build, and maintain production-grade LLM systems;
Own the full LLMOps lifecycle: data preparation, fine-tuning, evaluation, deployment, monitoring, and iteration;
Build evaluation frameworks to track quality, robustness, and regressions;
Work closely with researchers and engineers to productionize research prototypes;
Contribute to architecture decisions for agentic systems, RAG pipelines, and hybrid ML + symbolic solutions.
What We’re Looking For
5+ years of experience in ML / Applied AI;
Hands-on experience deploying ML or LLM systems into production;
Practical experience with LLMs (open-source or proprietary) in real use cases;
Strong understanding of modern LLM stacks: fine-tuning, inference optimization, RAG, agents;
Experience with MLOps / LLMOps tooling (evaluation, monitoring, CI/CD for models);
Strong Python skills; familiarity with ML frameworks;
Experience with Docker, Kubernetes (or equivalents) and GPU environments;
Startup or R&D-heavy background is a strong plus;
Fluent conversational English — you’ll work daily in an international team.
What’s Offered
16–22M JPY gross per year (~$105k–145k gross);
High autonomy, low bureaucracy, real ownership;
English & Japanese language learning support;
Full Japanese social security + additional medical coverage;