With the fast increase in computational power over the last decade, including the development of better Graphical Processing Units (GPUs), the field of Machine Learning (ML) and Artificial Intelligence (AI) has improved drastically since its inception in the 1980s. With more efficient algorithms and faster training times, the prominence of ML/AI throughout business and technology sectors has only grown. However, its extension to Nuclear Physics and Engineering remains limited to date. In this work, we explore the possible roles of ML in the Nuclear Data Evaluation pipeline including the development of (1) NucML and (2) an ML-based solution for neutron-induced evaluations. To catalyze future development in the area, NucML, the first end-to-end ML-augmented nuclear data evaluation pipeline was developed. This Python toolbox contains various capabilities ranging from loading ML-ready datasets to automatic validation of trained models using an in-house developed criticality benchmark public repository. It allows any user to focus on model development, in addition, to quickly navigating each step of the proposed enhanced pipeline in a modular easy-to-use fashion. Using the ML-enhanced framework and the newly developed code, several proof-of-concept ML models including Decision Tree, K-Nearest-Neighbor, and Gradient Boosting Machines were fitted to nuclear data from the EXFOR database to infer neutron-induce reaction cross sections. All models were used to predict nuclear data for several well-characterized isotopes in literature including Pu-239, U-238, U-235, and U-233. Afterward, data for several isotopes of interest including Cl-35, a less studied but important nuclide for advanced nuclear reactors, was generated. The predicted values for several of these isotopes were used to create an evaluation and tested using various benchmarks using SERPENT2 including the U-233 Jezebel and U-235-reflected U-233 spheres. Results for these proof-of-concept models in some cases outperforming the current release of the ENDF library by up to 200%. Once the models were validated, they were used to predict the Cl-35(n,p)S-35 at the same energy points as the new LBNL/UCB measurements. The predicted values described the cross sections 150% more accurately than any of the evaluated data libraries, which overestimate experimental results by up to a factor of five. The proof-of-concept models explored in this work, reliant on learning underlying patterns of cross section data from other radionuclides, demonstrate evidence that ML models can aid traditional physics-guided models and have a role to play in nuclear data evaluations. Furthermore, incorporating ML models in the nuclear data pipeline can allow evaluators to make faster human-bias-free decisions in areas of uncertainty as well as better inform future data measurement campaigns on areas of greatest sensitivity in EXFOR.