Generative AI Engineer

About the role

Blue Polaris is looking for a Generative AI Engineer to design, build and train custom AI and ML models for its Latin American engagements. The role is hands‑on and will own the full lifecycle of bespoke neural‑network solutions, from architecture design to production deployment.

Key responsibilities

• Design and implement custom neural‑network architectures, including novel layers, attention mechanisms and transformer components.
• Develop and train a variety of model types such as encoder‑only, encoder‑decoder, decoder‑only, CNNs, RNNs, Vision Transformers and diffusion models.
• Lead end‑to‑end model development: data preparation, large‑scale training, evaluation and iterative improvement.
• Apply advanced training techniques like reinforcement learning (PPO, RLHF, DPO, SAC) and parameter‑efficient fine‑tuning (LoRA, QLoRA), quantization, distillation and distributed training.
• Integrate models into production pipelines, ensuring scalability, reproducibility and robust MLOps practices.
• Collaborate with cross‑functional teams and stay up‑to‑date with emerging generative AI and multimodal architectures.

Required profile

• Bachelor’s, Master’s or PhD in Computer Science, AI, ML or a related field.
• 3–7 years of hands‑on deep‑learning experience with a proven record of building custom neural networks.
• Experience delivering production‑grade AI systems at scale.

Required skills

• Python (expert level)
• PyTorch and/or TensorFlow
• Hugging Face Transformers ecosystem
• Distributed training frameworks
• Transformer architectures and attention mechanisms
• Reinforcement learning (PPO, RLHF, DPO, SAC)
• Model optimisation techniques (LoRA, QLoRA, quantisation, distillation)
• MLOps tools for versioning, monitoring and retraining pipelines
• Version control (Git) and testing best practices
• Cloud training infrastructure (AWS, Azure, GCP)