Deep Learning Engineer

Responsibilities:

• Train and fine-tune deep learning models using hands-on model development practices.
• Work on pre-training, continued pre-training, supervised fine-tuning (SFT), PEFT, LoRA, QLoRA, distillation, RLHF, RLAIF, self-supervised learning (SSL), and domain adaptation.
• Develop and optimize modern Transformer and foundation models, including GPT-style models, LLaMA, Qwen, Mistral, and Gemma.
• Build and adapt vision-language models such as Qwen-VL and LLaVA.
• Work with Vision Transformers (ViT), LayoutLM, document transformers, and speech/audio transformers.
• Design custom Transformer architectures.
• Deploy models to production with high-throughput inference and GPU optimization.
• Support distributed training and inference workflows.
• Monitor and evaluate model performance in production environments.

Requirements:

• Strong hands-on experience in model training and fine-tuning of deep learning models.
• Experience with PyTorch, TensorFlow, or JAX.
• Strong understanding of Transformers, CNNs, ViTs, attention mechanisms, and representation learning.
• Experience in fine-tuning, optimization, distributed training, and model evaluation.
• Experience with production model deployment and ML infrastructure.
• Knowledge of high-throughput inference, GPU optimization, and tools such as vLLM, TensorRT, Triton, and SGLang.
• Experience with quantization, pruning, and distillation.
• Familiarity with Kubernetes and model monitoring.
• Nice to have: Experience with modern foundation models across language, vision-language, document, and speech/audio domains.