Genome assembly is the process of reconstructing entire DNA sequences from fragmentary input. While high-throughput sequencing technology has revolutionized the field of genomic science, assembly, often the first step of many bioinformatics pipelines, remains a computationally expensive problem which is further compounded by the difficulty to evaluate its accuracy. In this talk, I will describe the assembly algorithms, applications and practices. I will focus on a parallel tool we have developed for metagenomic assembly and discuss how it could be improved by building on a kmer-matching library. Finally, I will touch on an extensible pipeline service we are developing to help users find the best assembly.