Diffuse midline glioma (DMG) is a paediatric brain cancer characterised by infiltrative growth, resistance to treatment, and a median overall survival of 9–11 months. Progress towards effective therapies is limited by the availability of experimental models that reproduce the human developmental context in which these tumours arise and the cellular interactions that influence their behaviour. This thesis develops a human stem cell-based modelling framework that connects brainstem development and tumour initiation with immune–tumour interactions and the engineering of vascularised microenvironments. Human embryonic stem cells were differentiated into brainstem organoids (BrOs) through an FGF4-driven patterning strategy recapitulating aspects of early pontine development. Introduction of oncogenic mutations, including H3.3K27M, generated DMG organoids (DMGOs) exhibiting characteristic features of the disease, including infiltrative growth and intratumoural heterogeneity. Brainstem–pontine glial specification was critical for tumorigenesis, identifying regional developmental identity as a determinant of tumour formation in this model. These findings establish developmental context as an essential consideration in constructing biologically relevant human models of DMG. Incorporation of microglia and prolonged exposure to GD2 CAR T cells extended the platform to investigate immune–tumour interactions and treatment responses. The model captured clinically relevant variability in T-cell responses and demonstrated that tumour-associated microglia can diminish treatment efficacy. These observations highlight the contribution of the immune microenvironment to therapeutic response and provide a framework for investigating immunosuppressive interactions in a human brainstem tumour context. Reproducing this cellular complexity requires engineered environments that accommodate the mechanical, biochemical, and transport requirements of neural tissue. A review of neural tissue bioengineering therefore informed the integration of stem cell biology with bioprinting and microfluidic approaches. To make these technologies more accessible, an off-the-shelf mSLA 3D printer was converted into AnyBio, an open-source bioprinter assembled for approximately €320. The platform supports high-resolution fabrication of microfluidic structures using photocrosslinkable hydrogels, including GelMA and PEGDA. Evaluation of vitamin C as a radical scavenger, alongside photoabsorbers, addressed the balance between print fidelity and cell compatibility. To investigate vascular integration, BrOs were combined with naïve human endothelial cells, which acquired blood–brain barrier-like transcriptional signatures, although vascular penetration remained largely peripheral. Integration into bioprinted microfluidic chips used a mechanically tuned hydrogel approximating native brain stiffness. A custom bioreactor demonstrated the feasibility of directional laminar flow within physiological shear stress ranges. These advances establish components of a perfusable brainstem model, while sustained perfusion and full vascular coupling remain unresolved challenges. Endothelial transcriptional changes likewise require functional validation before barrier properties can be established. Together, this work positions regional developmental identity and microenvironmental complexity as complementary foundations for modelling DMG. It establishes a tractable human platform for studying tumour initiation and immune modulation, alongside accessible engineering tools for developing more physiologically representative culture systems. The resulting framework supports mechanistic investigation and provides a basis for further development towards predictive preclinical testing. By defining both demonstrated capabilities and outstanding biological and technical requirements, this thesis advances the development of human-relevant models with the potential to reduce reliance on animal experimentation in research on lethal paediatric brain cancer.