Cross-lingual transfer learning enables a promising solution to the paucity of annotated data in low-resource languages by facilitating syntactic knowledge transfer from high-resource languages. In this paper, we introduce the cross-lingual transfer learning paradigm for learning dependency parsers and part-of-speech(POS) taggers on one or more source languages and generalizing to the use on multiple target languages. We utilize multilingual embeddings, pre-emptive language-agnostic features, and fine-tuning techniques to facilitate generalization. Experiments with Universal Dependencies(UD) treebanks show sig- Meaningful enhancements in parsing and tagging per- formance, particularly for low-resource languages, thus lowering annotation cost as well as enhancing the scalability of multilingual natural language processing (NLP) systems.