How would you merge large quantities of Python-accessible information into a pre-existing set of XML documents.  This approach combined lxml.etree (ElementTree) with Kid templates (based on ElementTree) to efficiently annotate a large tree of documents.  We'll look at how to use ElementTree (and how to use the query mechanism in lxml's implementation to search your ElementTree trees) and how to quickly generate XML from your Python objects using Kid so that the results can merge into the pre-existing XML.

[Update] Presentation is online.