Aim Intelligence
Solutions Engineer
Seoul, South KoreaPosted today
What you'd do
- Define AI safety and reliability metrics quantitatively and drive practical improvements
- Design and run experiments combining SFT,RL,prompting,and representation steering techniques
- Build complete experiment loops for continuous product-level performance improvement
What they want
- Advanced Python for research-grade code and library-level implementation
- Hands-on experience with Hugging Face Transformers,PEFT,and model training ecosystems (LLM,VLM)
- Practical experience with RL,DPO,or other model alignment techniques
Nice to have
- PyTorch,JAX,or other deep learning framework experience
- CUDA/GPU kernel programming or high-performance inference optimization
- Distributed training frameworks (DeepSpeed,FSDP,Megatron-LM)