HTML5 Websockets Video/Audio Conference Module Expert

We are in need of a advanced web conferencing system with the following features: Core Features: 1. One to one audio+video+text chat with file attachment. Example: a Skype video call. 2. One to many audio+video+text conference with file attachment. Possible to configure a host who have control of the conference. Also possible to configure only one person can stream the video, others can only use audio, Example: a virtual lecture/classroom. 3. One to many video+audio streaming where one person streams to hundreds of people and other user can only use text chat. Example. twitch TV, Youtube live streaming. Server Requirement: 1. Able to save video/audio/text history on the server. 2. Runs on linux or windows. 3. Highly scalable and can handle large number of online users on a single server. 4. Able to host multiple sessions and mixture of different type of sessions. 5. Highly configurable API interface to manage sessions/users/authentications etc. Client requirement: 1. Web based. 2. Maximum browser compatibility, runs in 90%+ web browsers 3. Minimum flash, most interface should be in CSS/HTML so that it can be modified easily. Flash only used to access webcam/mic and only used to display video frame. We had a look at openfire/redfire but is unsure how to even customize it meet the requirement. We are looking for an expert to guide us on implementing a solution to meet the above requirements, we don't care how it is achieved whether using proprietary or open source software as long as the feature matches what we need. We are not interesting in developing such system from scratch, we just need someone we can consult to piece together things that are readily available to complete the system.