Algorithm Engineer – LLM & AI Companion

Position Description

We are seeking an Algorithm Engineer to work on AI product development based on Large Language Models (LLMs).

This is a full-time R&D role that requires strong ownership of algorithmic solutions, covering problem definition, system design, and production deployment in real-world environments.

This role will support both:

• B2B client projects (enterprise use cases, private deployment, constrained environments)
• In-house C-end AI products, especially AI companion applications with long-term interaction

The focus of this role is not academic model research , but building reliable, scalable AI systems under real-world constraints such as limited model capability, cost, and latency.

Job Responsibilities

• Optimize and develop LLM-based applications using open-source models (e.g. Qwen, LLaMA, DeepSeek), including instruction tuning, inference optimization, and performance evaluation
• Improve usability, stability, and controllability of models in real-world scenarios via prompt engineering, RAG, and rule/system-based strategies
• Design and implement core algorithms for AI companion products, including:
  • Persona consistency and persona drift control
  • Long-term, multi-turn dialogue logic with state persistence
  • Basic emotion or state modeling with a focus on engineering robustness rather than academic accuracy
• Package algorithmic capabilities into scalable, production-ready modules usable across products and projects
• Work closely with product and engineering teams to deliver solutions under real-world constraints and incomplete requirements
• Participate in algorithm design, data processing, validation, and system optimization

Job Requirements

• Bachelor’s, Master’s, or PhD degree in a STEM-related discipline
• Hands-on experience applying LLMs or AI systems in real-world or production-like environments
• Ability to independently design, implement, and validate algorithmic solutions
• Strong proficiency in Python with solid engineering and debugging skills
• Practical understanding of working under constraints such as limited model capability, compute cost, and changing requirements
• Eligible to legally work in Hong Kong and commit to full-time R&D work

Preferred (not mandatory):

• Experience with AI companions, dialogue systems, agents, or virtual characters
• Experience delivering AI solutions for enterprise clients or private deployments
• Multimodal experience (image, audio, or video)
• Awareness of compute cost, system stability, and engineering trade-offs

This role is not focused on pure model research or academic benchmarking, but on building AI systems that can be reliably used by real users and enterprise clients over time.