Samsung Knox is seeking to develop on-device enterprise AI agents and frameworks to protect businesses and consumers using Samsung devices, and to define the future role of smartphones in business
Requirements
- Hands-on experience with LLMs and NLP applications, especially in enterprise or security contexts
- Strong knowledge of AI agent architectures, multi-agent coordination, and decision-making frameworks
- Experience building or optimizing retrieval systems (knowledge graphs, vector search, hybrid search)
- Familiarity with RAG pipelines and their integration with LLMs for enterprise applications
- Proficiency in post-training techniques (alignment, safety tuning, RLHF, evaluation)
- Experience with model optimization (quantization, pruning, distillation) for on-device/edge systems
- Strong coding skills in Python with frameworks such as PyTorch or TensorFlow
Responsibilities
- Explore and design novel agent architectures for enterprise-grade AI agents that combine automation, context-awareness, and secure decision-making
- Research and develop on-device RAG pipelines, integrating knowledge graphs, vector search, and hybrid keyword search to enable secure and contextual enterprise AI
- Optimize retrieval and reasoning workflows for agents, ensuring efficient integration of enterprise knowledge into LLM-based decision-making
- Conduct post-training optimization (safety tuning, alignment, and evaluation) for robust and trustworthy deployment of enterprise AI agents
- Apply model optimization techniques (quantization, distillation, pruning) for efficient deployment on mobile and IoT devices
- Build frameworks leveraging LiteRT LM, Mediapipe LLM, llama.cpp, MLC LLM, and similar runtimes for on-device agent deployment
- Rapidly prototype and develop proof-of-concept applications to explore new features and validate technical feasibility
Other
- Master's or Ph.D. in Computer Science, AI/ML or a related field, or equivalent combination of education, training, and experience
- 1+ years of experience
- Strong communication and teamwork skills, with a results-oriented approach
- Must not disclose any trade secrets of a current or previous employer
- Ability to work in an office setting and operate standard office equipment
