diff --git a/src/qinfer/parallel.py b/src/qinfer/parallel.py
index 7ccf820..2f5c269 100644
--- a/src/qinfer/parallel.py
+++ b/src/qinfer/parallel.py
@@ -68,6 +68,10 @@ def expparams_dtype(self):
     @property
     def is_n_outcomes_constant(self):
         return self._serial_model.is_n_outcomes_constant
+
+    @property
+    def n_engines(self):
+        return len(self._dv)
     
     ## METHODS ##
     
@@ -80,12 +84,24 @@ def n_outcomes(self, expparams):
     def likelihood(self, outcomes, modelparams, expparams):
         # By calling the superclass implementation, we can consolidate
         # call counting there.
-        super(DirectViewParallelizedModel, self).likelihood(outcomes, modelparams, expparams)
-        
+        super(DirectViewParallelizedModel, self).likelihood(outcomes, modelparams, expparams) 
+
+        # Need to decorate with interactive to overcome namespace issues with
+        # remote engines.
+        @IPython.parallel.interactive
+        def serial_likelihood(mps, sm, os, eps):
+            return sm.likelihood(os, mps, eps)
+
+        # TODO: check whether there's a better way to pass the extra parameters
+        # that doesn't use so much memory.
+        # The trick is that serial_likelihood will be pickled, so we need to be
+        # careful about closures.
         L = self._dv.map_sync(
-            lambda mps, sm=self._serial_model, os=outcomes, eps=expparams:
-            sm.likelihood(os, mps, eps),
-            np.array_split(modelparams, len(self._dv), axis=0)            
+            serial_likelihood,
+            np.array_split(modelparams, self.n_engines, axis=0),
+            [self._serial_model] * self.n_engines,
+            [outcomes] * self.n_engines,
+            [expparams] * self.n_engines,
         )
         return np.concatenate(L, axis=1)