The study presents a novel approach to engineer amaryllidaceae enzymes for biomanufacturing of therapeutic alkaloids, focusing on the Alzheimer's medication galantamine. The authors developed a biosensor-machine learning technology stack to improve the efficiency of biocatalyst development. They evolved the RamR transcription factor into a highly sensitive and specific biosensor for 4′-O-methylnorbelladine, a key intermediate in the amaryllidaceae alkaloid pathway. This biosensor was then used to monitor the activity of norbelladine 4′-O-methyltransferase (Nb4OMT) from *Narcissus pseudonarcissus* in *Escherichia coli*. A structure-based residual neural network, MutComputeX, was developed to generate activity-enriched variants of Nb4OMT, which were rapidly screened using the biosensor. The best variant showed a 60% improvement in product titer, 2-fold higher catalytic activity, and 3-fold lower off-product formation. The crystal structure of this engineered enzyme was solved, elucidating the mechanism behind the beneficial mutations. The study highlights the potential of biosensors and machine learning in accelerating the development of microbial fermentation processes for therapeutic alkaloids.The study presents a novel approach to engineer amaryllidaceae enzymes for biomanufacturing of therapeutic alkaloids, focusing on the Alzheimer's medication galantamine. The authors developed a biosensor-machine learning technology stack to improve the efficiency of biocatalyst development. They evolved the RamR transcription factor into a highly sensitive and specific biosensor for 4′-O-methylnorbelladine, a key intermediate in the amaryllidaceae alkaloid pathway. This biosensor was then used to monitor the activity of norbelladine 4′-O-methyltransferase (Nb4OMT) from *Narcissus pseudonarcissus* in *Escherichia coli*. A structure-based residual neural network, MutComputeX, was developed to generate activity-enriched variants of Nb4OMT, which were rapidly screened using the biosensor. The best variant showed a 60% improvement in product titer, 2-fold higher catalytic activity, and 3-fold lower off-product formation. The crystal structure of this engineered enzyme was solved, elucidating the mechanism behind the beneficial mutations. The study highlights the potential of biosensors and machine learning in accelerating the development of microbial fermentation processes for therapeutic alkaloids.