What You'll Do

• Fine-tune and deploy vision-language models (VLMs) and large language models for production use cases
• Design and maintain end-to-end pipelines for multi-modal model training, evaluation, and inference in Python
• Develop prompt engineering strategies, RAG architectures, and other techniques to maximize model performance
• Evaluate model outputs systematically and build feedback loops for continuous improvement
• Quantize large transformer models to improve model efficiency
• Stay current with rapid advances in transformer architectures, fine-tuning methods, and multi-modal research

Requirements

• 3+ years of professional experience working with transformer-based architectures
• 2+ years of hands-on experience fine-tuning and deploying multi-modal models (VLMs)
• 2+ years of proven computer vision experience, with a strong preference for object detection
• Strong experience with LLMs — fine-tuning, inference optimization, and production deployment
• Proficiency in Python for model development, training, and deployment (2+ years)
• Experience with deep learning frameworks such as PyTorch or TensorFlow
• Solid understanding of attention mechanisms, tokenization, transfer learning, and generative model fundamentals
• Proven experience taking models from experimentation through production-ready deployment

Nice to Have

• SQL proficiency for querying detection results, labeling metrics, or model performance data
• Strong preference: experience with roadside or infrastructure object detection (signs, signals, debris, pavement markings)
• Background in GovTech, public sector, or smart city projects
• Experience in automated driving, ADAS, or autonomous vehicle perception systems
• Familiarity with model-assisted labeling, active learning, or human-in-the-loop workflows
• Experience with edge deployment or model optimization (TensorRT, ONNX, quantization)