Autoencoders are unsupervised class of neural networks which gained their prominence in modern Artificial Intelligence (AI) era due to their versatility in data compression, future extraction and generative modeling. In recent years, autoencoders have evolved as an indispensable tool, driving advancements in fields such as data analysis, image processing, and AI-powered automation. These deep neural networks are great for tasks like noise reduction, anomaly detection, and feature extraction because they can reflect the most important parts of a dataset by compressing the raw input data. Over the time, different variants of autoencoders have been evolved to address the limitations of traditional autoencoder models. One such variant is the Variational Autoencoder (VAE), which is highly preferred for its enhanced probabilistic approach. This distinguishes them from the traditional autoencoders that use the deterministic mapping from input to latent representation. Modeling the latent variables using the probability distributions will help in generating new data points by taking samples from latent space. This breakthrough makes the VAEs exceptionally a powerful generative model, frequently applied in fields such as image generation, anomaly detection, and semi-supervised learning. Their capacity to interpolate between data points and produce realistic examples which makes them ideal for tasks that demand both creativity and precision in predictions.