Lead AI/ML Engineer

**About the company:** Starley Minato-ku, Tokyo At Starley, we’re developing "Cotomo," a leading voice-based AI conversation platform. Our goal is to redefine how humans interact with AI. Read more **Responsibilities:** Own Cotomo’s ML stack end-to-end across language, speech, image, video, and multimodal models Improve conversational quality, personality, memory, personalization, latency, reliability, and cost Build and improve training, post-training, evaluation, and experimentation pipelines Evaluate and A/B test frontier APIs, open-weight models, and in-house models, making pragmatic decisions based on product quality and unit economics Own production ML systems, including model serving, GPU infrastructure, observability, and cost optimization Work closely with Product and Engineering to turn new model capabilities into measurable user impact Provide technical direction, reviews, and hands-on support for our small ML team Requirements 4+ years of full-time professional experience in machine learning, AI, or a closely related engineering role You’ve built and operated production ML systems and are comfortable owning them beyond the experimentation stage You have a strong understanding of modern deep learning and LLM systems across training, fine-tuning, evaluation, and inference Strong software engineering skills come naturally to you, including working with GPU-based ML infrastructure Ambiguous technical problems don’t need to be fully scoped before you start — you can define the problem, make trade-offs, and drive it through to production You think in terms of product outcomes, not just model benchmarks You’re pragmatic about architecture and are equally comfortable choosing an external API, an open-weight model, or an in-house solution when it is the best fit You can set technical direction and raise the bar for a small team while remaining primarily a hands-on individual contributor Nice to haves While not specifically required, tell us if you have any of the following. Conversational AI, AI companions, character AI, voice AI, or other consumer AI products ASR, TTS, image generation, video generation, or other multimodal systems Post-training, preference optimization, reinforcement learning, synthetic data, or personalization Distributed training or large-scale inference systems Latency, reliability, and cost optimization for production ML Staff, Principal, Lead, or senior Research/ML Engineer-level technical ownership Compensation ¥9,000,000 ~ annually. With performance-based stock options. **Requirements:** 4+ years of full-time professional experience in machine learning, AI, or a closely related engineering role You’ve built and operated production ML systems and are comfortable owning them beyond the experimentation stage You have a strong understanding of modern deep learning and LLM systems across training, fine-tuning, evaluation, and inference Strong software engineering skills come naturally to you, including working with GPU-based ML infrastructure Ambiguous technical problems don’t need to be fully scoped before you start — you can define the problem, make trade-offs, and drive it through to production You think in terms of product outcomes, not just model benchmarks You’re pragmatic about architecture and are equally comfortable choosing an external API, an open-weight model, or an in-house solution when it is the best fit You can set technical direction and raise the bar for a small team while remaining primarily a hands-on individual contributor **Nice to have:** While not specifically required, tell us if you have any of the following. Conversational AI, AI companions, character AI, voice AI, or other consumer AI products ASR, TTS, image generation, video generation, or other multimodal systems Post-training, preference optimization, reinforcement learning, synthetic data, or personalization Distributed training or large-scale inference systems Latency, reliability, and cost optimization for production ML Staff, Principal, Lead, or senior Research/ML Engineer-level technical ownership **Compensation:** ¥9,000,000 ~ annually. With performance-based stock options.