Direct bioproduction of DHAA (dihydroartemisinic acid) rather than AA (artemisinic acid), as suggested by previous work would decrease the cost of semi-biosynthesis artemisinin by eliminating the step of initial hydrogenation of AA. The major challenge in microbial production of DHAA is how to efficiently manipulate consecutive key enzymes ADH1 (artemisinic alcohol dehydrogenase), DBR2 [artemisinic aldehyde Δ11(13) reductase] and ALDH1 (aldehyde dehydrogenase) to redirect metabolic flux and elevate the ratio of DHAA to AA (artemisinic acid). Herein, DHAA biosynthesis was achieved in by introducing a series of heterologous enzymes: ADS (amorpha-4,11-diene synthase), CYP71AV1 (amorphadiene oxidase), ADH1, DBR2 and ALDH1, obtaining initial DHAA/AA ratio at 2.53. The flux toward DHAA was enhanced by pairing fusion proteins DBR2-ADH1 and DBR2-ALDH1, leading to 1.75-fold increase in DHAA/AA ratio (to 6.97). Moreover, to promote the substrate preference of ALDH1 to dihydroartemisinic aldehyde (the intermediate for DHAA synthesis) over artemisinic aldehyde (the intermediate for AA synthesis), two rational engineering strategies, including downsizing the active pocket and enhancing the stability of enzyme/cofactor complex, were proposed to engineer ALDH1. It was found that the mutant H194R, which showed better stability of the enzyme/NAD complex, obtained the highest DHAA to AA ratio at 3.73 among all the mutations. Then the mutant H194R was incorporated into above rebuilt fusion proteins, resulting in the highest ratio of DHAA to AA (10.05). Subsequently, the highest DHAA reported titer of 1.70 g/L (DHAA/AA ratio of 9.84) was achieved through 5 L bioreactor fermentation. The study highlights the synergy of metabolic engineering and protein engineering in metabolic flux redirection to get the most efficient product to the chemical process, and simplified downstream conversion process.