I noticed, that as soon as a lambda field is involved in a query returning a list of results, the queries take multiple seconds to complete. I suspect that dgraph is performing a single lambda request for each field. I would suggest, if it’s feasible, to batch request lambda queries. As far as I can tell it’s not possible to nest lambda fields. (Although this is technically possible to define in the schema, it won’t actually work). This would mean the query could be executed and only after that the lambda fields would have to be resolved, so this sounds like something that could be batched.
Filing this under performance. @hardik please do take note