Our expertise is in an automated design of highly optimized digital circuits such as arithmetic circuits, signal processing primitives, image operators, cryptography relevant functions, etc. The methodology is based on efficient parallel implementations of multi-objective evolutionary algorithms such as genetic programming. We are capable of developing not only optimized exact implementations of these circuits but also their approximate implementations. High-quality trade-offs among the error, power consumption and other electrical parameters are routinely obtained in the so-called functional approximation scenario. A library of approximate arithmetic circuits is available online http://www.fit.vutbr.cz/research/groups/ehw/approxlib/