<p>In this article we consider the classical inference of a semi-parametric stage life testing model under time constraint. We assume only two stress levels of a stage life testing experiment and a simple step-stress life testing experiment becomes a special case under this set up. We do not assume any specific parametric form of the lifetime of experimental units, rather we assume a piece wise increasing, decreasing or a constant hazard rate for the experimental units. The Weibull distribution becomes a special case under this model assumption. It is well known that due to its flexibility the Weibull distribution is a widely used lifetime distribution for analyzing time to event data. The proposed semi-parametric model is more flexible than the Weibull model and based on the data obtained from a stage life testing experiment we assume the piece wise increasing or decreasing or a constant hazard rate function. We have obtained the maximum likelihood estimators of the model parameters. Though the model involves significant number of unknown parameters, we just need to solve one or two dimensional optimization problem. Since the small sample properties of the maximum likelihood estimators are difficult to obtain we propose to use asymptotic properties for the construction of the confidence intervals of the unknown parameters. An extensive simulation study has been performed to assess the performance of the proposed estimator. One simulated data from stage life testing experiment and one real data from step-stress life testing experiment have been analyzed for illustrative purpose. In real data analysis, it has been observed that, proposed semi-parametric model fits the data better than the Weibull model.</p>