Heterogeneity in recombinant protein expression is a critical challenge in baculovirus expression vector system (BEVS) bioprocessing, yet many platforms lack an analytical approach that can simultaneously report infection status, endogenous biochemical state, and cell morphology on individual cells. We present an automated multimodal platform combining high-throughput Raman spectroscopy, phase-contrast microscopy, and fluorescence imaging through a single high-NA objective, acquiring coregistered spectra and images from approximately 1000 individual Spodoptera frugiperda (Sf) 9 cells in ∼32 min. An integrated data-processing pipeline links phase-contrast–derived morphological features, fluorescence-validated reporter abundance, and label-free Raman spectra at single-cell resolution. Applied to three BEVS conditions, i.e. noninfected cells, cells expressing mCherry (BC), and cells coexpressing mCherry with a GLUT6-related transporter (B8), a PCA–LDA model trained on Raman spectra achieved 93% cross-validation accuracy. Cross-modal analysis shows Raman spectroscopy captures biochemical information inaccessible to fluorescence. The mCherry-associated Raman band correlated strongly with fluorescence in BC (Spearman ρ = 0.83) but not in B8 (ρ = −0.16), indicating that Raman captures construct-dependent biochemical states that complement fluorescence-based reporter measurements. Phase-contrast–derived radiomic features independently corroborated this, with amplified cell swelling in high-expressing B8 cells relative to BC (Cliff’s δ = −0.69 versus −0.51) pointing to an additional biosynthetic burden in the dual-expression sample. By resolving construct-specific biochemical, morphological, and expression heterogeneity at single-cell resolution, this multimodal workflow demonstrates the analytical potential of correlated single-cell measurements for BEVS characterization, establishing an analytical basis for future at-line adaptation.