ML Engineer at an AI companion startup

About The Job

The client is hiring for this role. Harper is an AI career agent that works for you — the candidate. Skip the application form: just talk to Harper, and it puts strong candidates straight in front of the client — plus other great opportunities worth your time.

Job Title

ML Engineer

Salary

Competitive base + Equity — early-team package at a fast-growing AI startup

Location

Seoul (Gangnam)

Company Description

Founded by the team behind Hyperconnect — whose video app Azar was acquired by Match Group for ~$1.7B (₩2 trillion) — this is a fast-growing startup building real-time AI companions: lifelike AI characters you can see, hear, and talk to over live video, not just text chat. A proven consumer-social team now going all-in on AI entertainment for a global audience, with a major push into North America. Seoul-based, and early enough that you'll shape both the product and the tech.

About The Role

You'll build the core ML that lets AI companions talk, react, and connect with users in real time — not model research for its own sake, but production-grade real-time video, voice, and conversation.

What You'll Do

• Generate and play back real-time video with natural lip-sync and motion driven by user speech and context — interactive avatars / talking-head / lip-sync.
• Build the end-to-end real-time pipeline that connects video, voice, and conversation models.
• Optimize latency, video quality, and cost for real production usage.
• Design conversation systems that make users want to keep talking — relationship-centered dialogue tuned for live voice/video, with long-term memory.
• Monitor conversation quality in production and keep improving it from real user behavior.

Who You Are

• 3+ years in AI/ML (including graduate studies), with 2+ years of hands-on industry experience.
• Strong AI/ML fundamentals, plus real depth in at least one domain.
• You've taken models beyond research — applied to real products and served in production.
• You care as much about how users actually feel as about model metrics.
• Fluent Korean communication.

Bonus

• Talking-head, lip-sync, or virtual human / avatar research or development.
• Flow-based / diffusion image and video generation.
• Voice conversation systems (STT / LLM / TTS, or end-to-end models).
• Persona-based, entertainment, or long-term-memory chatbot design.
• GenAI B2B/B2C product experience; ML microservices on cloud (containers, Kubernetes); ML-driven retention/engagement wins; top-venue publications.

Why Join

• Build alongside a proven team — the founders behind a global app acquired by Match Group for ~$1.7B.
• Build genuinely novel real-time video AI — one of the hardest, most exciting problems in applied ML.
• Ship to a global consumer audience and watch real users react.
• Early enough to shape the product and own the ML stack.
• Seoul-based, fast-moving team. Competitive comp + equity.