Passionate and dedicated Machine Learning Engineer, specializing in developing scalable ML solutions. Proficient in tackling complex challenges through efficient problem-solving. Experienced in both research and practical applications, with a strong focus on collaboration to drive innovative outcomes.

• Finetuned various VLM including NVILA, Qwen2.5VL, and LlAVA using GRPO on video understanding tasks, especially industrial aspects with 54% improvement especially on instance actions.
• Utilized temporal grounding model LITA for few shot video fine-tuning on industrial datasets using reinforcement learning policy optimization methods and gained the IOU with 35%.

• Developed a Convolutional-Recurrent model using C++ and Python for touchpad gesture and mouse movement recognition, and applied TensorRT to optimize and deploy the model to internal tools with real-time response.
• Achieved an impressive average accuracy of 98% with the proposed classification model with low memory usage.
• Leveraged Tensorflow's C++ API for end-to-end processes: from dataset curation, model design, model optimization, model deployment. The pipeline can be migrated to various scenarios for quick development.

• Integrated Azure Cognitive Service with PoC for cost-effectiveness and efficiency to meet customer needs.
• Finetuned Large-Language Models (Opt-models) utilizing the DeepSpeed framework on proprietary enterprise datasets, enhancing production efficiency for internal applications.
• Incorporated Azure OpenAI services (GPT-3.5, GPT-4) into Microsoft Teams and conducted finetuning using client- provided datasets, bolstering communication efficacy.

• Developed "Transfermer," a novel Transformer-based MARL framework achieving 50% improved training efficiency
• Integrated few-shot and zero-shot learning techniques into pre-trained Multi-agent Reinforcement Learning models

• Advanced brain signal simulation using GPT2 model, achieving 25% improved accuracy
• Developed innovative GPT2xCNN architecture for enhanced signal generation quality

"HGAP: Boosting Permutation Invariant and Permutation Equivariant in Multi-Agent Reinforcement Learning via Graph Attention Network" (ICML 2024, First Author)

• 🌱 I’m currently learning Multi-agent reinforcement learning, Robotic control and Deep learning

• 👨‍💻 All of my projects are available at https://github.com/ChristianLin0420

• 💬 Ask me about Artificial Intelligence

• 📫 How to reach me crlc112358@gmail.com

• 📄 Know about my experiences https://drive.google.com/file/d/1Wo7asiI1xl6x8zAvql2R1EpF1esoVMRu/view?usp=sharing

• Deep Learning Specialization (DeepLearning.AI)

• TensorFlow: Advanced Techniques Specialization (DeepLearning.AI)

• Getting Started with Accelerated Computing in CUDA C/C++ (Nvidia)

• Project Management Specialization (Google)

• Love to learn neuroscience in order to create a real AI!!!!