The PEAK1 and Pragmin/PEAK2 pseudo-kinases have emerged as important components of the protein tyrosine kinase pathway implicated in cancer progression. They can signal using a scaffolding mechanism that involves a conserved split helical dimerization (SHED) module. We recently identified PEAK3 as a novel member of this family based on structural homology; however, its signaling mechanism remains unclear. In this study, we found that, although it can self-associate, PEAK3 shows higher evolutionary divergence than PEAK1/2. Moreover, the PEAK3 protein is strongly expressed in human hematopoietic cells and is upregulated in acute myeloid leukemia. Functionally, PEAK3 overexpression in U2OS sarcoma cells enhanced their growth and migratory properties, while its silencing in THP1 leukemic cells reduced these effects. Importantly, an intact SHED module was required for these PEAK3 oncogenic activities. Mechanistically, through a phosphokinase survey, we identified PEAK3 as a novel inducer of AKT signaling, independent of growth-factor stimulation. Then, proteomic analyses revealed that PEAK3 interacts with the signaling proteins GRB2 and ASAP1/2 and the protein kinase PYK2, and that these interactions require the SHED domain. Moreover, PEAK3 activated PYK2, which promoted PEAK3 tyrosine phosphorylation, its association with GRB2 and ASAP1, and AKT signaling. Thus, the PEAK1-3 pseudo-kinases may use a conserved SHED-dependent mechanism to activate specific signaling proteins to promote oncogenesis.