Akshay Kanade

• Designed and deployed deep-learning models for detection, segmentation, and multimodal vision tasks.
• Implemented model explainability methods to improve transparency and trust in AI decisions.
• Optimized GPU utilization and inference latency through mixed-precision training, ONNX conversion, and Tensor RT
• Built automated ML pipelines using MLflow for training, evaluation, and CI/CD deployment.
• Led experimentation and benchmarking of state-of-the-art foundation models to identify best-performing architectures for diverse computer vision tasks.

As an AI/ML Engineer at OneVision, I lead the development of generative AI pipelines, prompt-engineering frameworks, LLM fine-tuning, multimodal model evaluation, and RAG-based architectures. I have implemented reusable agent workflows, optimized inference pipelines using TensorRT/ONNX, and improved prompt-template relevance by 40%. My work spans model lifecycle management, GPU optimization, CI/CD automation, and production monitoring—mirroring your expectations for enabling AI multipliers, building scalable templates, and ensuring high-quality technical operation.

I have completed master in Electrical and embedded system form Ravensburg and weingarten university.