Developing efficient microbial processes to produce flavonoids has been a metabolic engineering goal over the past decade due to their important functions. Previously, the de novo production of the main flavonoid precursor (2S)-pinocembrin was achieved. However, low productivity and two separate fermentation steps made it inappropriate for industrial scale low-cost production. Here, a stepwise modular engineering approach was introduced to systematically identify and eliminate metabolic pathway bottlenecks. The overall pathway was firstly divided into four modules and analysis then revealed that efficient conversion of l-phenylalanine to (2S)-pinocembrin is the major limiting factor. Therefore, the pathway from l-phenylalanine to (2S)-pinocembrin was re-cast into three modules to alleviate this bottleneck by modifying both the gene copy number and promoter strength. Furthermore, the expression of the rate-limiting enzyme PAL was found to be correlated with 5' region of mRNA structure. The efficiency of the synthetic pathway was then improved by customizing the PAL expression level based on modification of the mRNA secondary structure. Fed-batch cultivation of engineered strains in a 3-L fermentor resulted in a final (2S)-pinocembrin production of 432.4mg/L. The results presented here pave the way for the development of an economical and simple process for microbial production of flavonoids.