Life-course research asks how employment and family trajectories shape long-run outcomes, yet most empirical studies rely on a cluster-then-regress strategy that cannot recover causal effects when treatment and covariates evolve jointly over time. I propose a two-stage framework that integrates sequence analysis with g-computation. In the first stage, multi-channel sequence analysis and clustering identify empirically prevalent trajectory types whose medoids serve as counterfactual interventions for which positivity is empirically defensible. In the second stage, the causal effect of each medoid trajectory is estimated via g-computation with forward simulation of time-varying covariates via SuperLearner. A semi-synthetic simulation calibrated to the NLSY79 reveals a clear estimator hierarchy: cluster-adjusted regression recovers only a small fraction of the true causal effect, while flexible g-computation recovers the true effects most accurately. An empirical application to NLSY79 data reveals substantial wealth effects of sustained full-time employment and stable family formation, with notable divergence between parametric and flexible estimators. The framework offers life-course researchers a practical approach for defining trajectory interventions and estimating their causal effects.