Instituto Butantan (a.k.a. the Snake Farm) described in Coffee Crash is a real biological research facility in São Paulo, Brazil. Among its many public health functions, the institute breeds snakes, spiders, and other poisonous critters in order to produce anti-venoms.

Although I didn't mention this in the book, a portion of the institute suffered a major fire in May 2010. Until that time, the institute had housed a collection of more than 80,000 preserved (dead) snake specimens, as well as nearly half a million preserved spiders and scorpions. The fire destroyed the entire collection, fueled by the highly flammable formaldehyde and ethanol used for specimen preservation. The fire also destroyed the institute's main research library, including its books. Fortunately, no people or living creatures were harmed in the fire, but the Snake Farm may never fully recover the  research potential of its collection lost in the fire.

More information on Instituto Butantan is available at http://www.butantan.gov.br/home/ (currently in Portuguese only).

A scene in Coffee Crash describes a method by which the U.S. National Security Agency could crack Secure Sockets Layer 128-bit encryption with the help of "an inverse adaptation of the Chinese Remainder Theorem," and by exploiting the fact that most computer generated random numbers are not truly random. (A massively distributed secret supercomputer system is also described. For more on that, see my Author's Blog post of 05/23/2012.)

The RSA public key/private key encryption method used for typical Internet applications does employ the Chinese Remainder Theorem to speed up decryption for a message's intended recipient, making the calculations about four times faster than would otherwise be the case. But the Chinese Remainder Theorem already uses inverse calculations. There's no such thing as an "inverse adaptation" of it. I just made that up.

It is true that most computer-generated random numbers are not truly random, just pseudorandom. The NSA undoubtedly knows ways in which these pseudorandom numbers differ from truly random, and could gain some advantage in attempting to decrypt an intercepted message. However, the advantage gained by such knowledge would still be quite small compared to the overall difficulty of cracking the encyption.

Unless some element of the original encryption was compromised at the outset, it's unlikely that the NSA can crack a 128-bit encrypted message in any reasonable time frame with today's technology. On the other hand, a March 2012 article by James Bamford in Wired magazine describes some of the NSA's forthcoming efforts in that realm.