Online social networks have seen a rapid increment in the number of users across the globe. The giant social networks like Facebook, Twitter, Reddit, Flicker etc do have billions of users and are increasing day by day. A recommender system is an engine which takes a particular user as an input and gives the preferential output which recommends the input user befriend with someone or join a group or get subscribed to an interest or to simply buy a product. To help users every site does have a recommender system. But these recommender systems are mostly based on the fact that the user friends have subscribed to the recommended page/group. In this work, we researched over the Reddit Social Network which is one of the major Social Networking Sites. We engineered a recommender system which recommends the users to follow the more specific Sub-Reddits feeds and is solely based on the current subscribed Sub-Reddits of a user instead of demographic knowledge or Friends knowledge. We used Link Prediction in Social Network as our base and used our novel approach of Weighted Adamic Adar score to find the Sub-Reddits which are closely related to the Sub-Reddits already linked with the user profile. In this way, we predicted the new Sub-Reddits in which the user might have interest. In this work, we addressed the issue of Cold Start for a user who does not have any prior 'like', 'interest' or item purchase history.