Is he a scientist or is he a philosopher? He’s a philosopher. His contributions to science (and anything else) are philosophical. He’s learned a great deal from science and is very pro-science. Some, but not so many, scientists appreciate philosophical input or explicitly concern themselves with the philosophical underpinnings of their subject.
Is he a “four horsemen” atheist or not? Yes, he is an atheist, but unlike most others, he’s not against religion per se, it has value in love and belonging (religiare – that which binds us culturally) and sees an entirely naturalistic explanation for religion. He’s against dogma and fanaticism – he certainly would “cage and/or disarm” extremists.
Does he believe consciousness is an illusion or not? No and Yes. It’s real and natural, but our typical view of it is illusory. It’s really a user interface behind which actual causality is quite different to the objects we can perceive in the interface. (How else would we / nature / god “intelligently-design” a supervisory dashboard for a complex system? – it’s a bag of tricks – The “mind” doesn’t need to know everything the “brain” is doing. See machine view.)
Does he explain or explain-away consciousness? Yes and Yes. Everything he has done has been of a piece attempting to explain and understand consciousness but, in “Consciousness Explained” and since, he has certainly been explaining away – pointing out an illusory idea – the “special” nature of consciousness – beyond natural evolutionary explanation. It’s a wonderful bag of natural tricks.
Does he believe in human exceptionalism? Yes and no. Human consciousness, intelligence and culture is the most wonderful bag of tricks we know about so far, but it’s no exclusion from natural evolution in any exceptional way.
Is he just a “materialist”, are minds just “machines”, can “culture” just be a product of the mind as materialist machine? Yes, but not “just”. Dawkins and Turing both showed us that evolution and computation are algorithmic, competence without comprehension, leading eventually to comprehension of competence. Robots made of robots made of robots made of robots made of robots made of …. Requires a stranger (loopy) version – an inversion – of reasoning more than mere greedy reductionism, reducing the “wonderful bag of tricks” to a direct chain of causation between the objects we choose to define (*). Always need to see big historical (evolutionary, fluid) picture as well as isolated local and temporal detail. [Post Note: ie Compatibilism is not simply (need not be) a fudge – after Julian Baggini on Freedom Regained. (*) This “shifting the basis of the argument” point is key to “Dennett’s wager“.]
Critics in both the Humanities and the Sciences need to shift their (appreciation of) materialist (naturalist) thinking away from that kind of objective reductionism. Neither need have anything to fear except dogmatism in what may or must be explained in any particular way.
Hat tip to Jim Al-Khalili and the “Life Scientific” of  Dan Dennett this morning.