When a crisis or a disaster occurs, first responders, news agencies and the victims often use social media to disseminate situational information. However, authorities involved in disaster management are still not comfortable in using social media data to take decision, since right kind of situational information in a consistent manner is difficult to extract from the unstructured and often unreliable social media posts. To get meaningful data from people at disaster site, we propose a system framework and a research prototype that uses interactive crowdsourcing using SMS and Twitter to collect data form the "crowd" at disaster site. We define crowdsourcing in this context as an adaptive process that relies on an interactive dialogue between our system and each member of the crowd (disaster victims/ volunteers/ first responders) willing to contribute consistent information that would help the system to build a structured repository of information. The same repository can be used to take decision and answer users' queries on specific issues related to that disaster. Experiments have been performed using both SMS and Twitter in the context of recent earthquake at Nepal to validate our system and demonstrate that interactive crowdsourcing is helpful in decision making to manage disaster.