Five of the six polls projected the BJP-led alliance mopping up the 272 seats needed to form a majority in the 543-member parliament. Predictions for the 28-party alliance have ranged from 249 to a whopping 340 seats.
All the polls projected the BJP to win more than 200 seats - between 210 and 291- besting its previous high of 182 seats in the 1998 and 1999 elections.
The surge in seats this time, the polls suggest, has been powered by massive gains in the politically crucial northern state of Uttar Pradesh which sends 80 MPs to parliament.
The polls are terrible news for the ruling Congress party: five of the six polls project the Congress slipping below three digits - between 57 and 85 seats.
If that is indeed the case when the votes are counted on Friday, it will be India's Grand Old Party's worst performance since 1998 when it got 114 seats.
"These two polls have created some confusion. The Times Now-ORG poll's projections, for example, for the main parties in states like Bihar and Rajasthan are difficult to believe."
Otherwise, Dr Kumar reckons, the exit polls this time are "going to be more or less accurate".
"I don't expect the projections to go wayward when the votes are counted," he says.
Opinion and exit polls have a mixed record in India, and in the recent past - 2004 and 2009 polls - have been way off the mark, underestimating the Congress-led alliance.
Questions have been raised in the past about the suspected manipulation of some polls by political parties, flawed methodology and unrepresentative sample sizes.
Such polls in India are based on random representative samples of voters drawn from the voter list of the country's Election Commission.
A truly wide and representative sample is critical to polling in a country where 70% of its people live in villages.
"That means any election survey in order to be representative should try to conduct about 70% of its interviews in the villages and rural areas and only about 30% interviews in the towns and cities. But this is not followed by most of the market research agencies that are engaged in election polling in India," writes Praveen Rai, a colleague of Kumar at the CSDS, in a paper on the subject.
Rai believes that the reasons for this "can be attributed to inconvenience and high costs of conducting surveys in rural areas".
Also, polling organisations, he says, don't spend time and invest enough in fieldwork training and practices for their enumerators and in making sure they reach out to a wider representative sample.
Already, investment bank Credit Suisse has warned that Monday's polls reveal a "range of high uncertainty". The bank said that while the national totals are within a relatively narrow range, the state seat numbers that add up to the grand total of 543 seats, varied "widely", possibly alluding to one of the "outlier" polls.
Imagine then, for a moment, a scenario where seats won by BJP dip to between 160 and 180 when the votes are counted on Friday. (The BJP alone has the lion's share of 116 of 141 MPs in its present 28-party alliance, and most of its current allies are smaller parties.)
That means they have to woo a number of additional - and potentially demanding - parties into the alliance who can bring them 92-112 seats required for a simple majority. History shows that the allies can be demanding, and that may even include a pitch for supporting a BJP-led government without Mr Modi at the helm, who many find a polarising figure.
Or imagine, another scenario, where the Congress party does better and wins more than 100 seats, and then tries to cobble together an alliance with the "other parties" which, according to the exit polls, are projected to win between 111 and 162 seats. Highly unlikely, but an obscure possibility nonetheless.
On the other hand, the polls may have even underestimated the BJP's march, in which case the BJP-led alliance with Modi at the helm would sail through with a simple majority.
Or, as Sanjay Kumar says, if five of the six polls have similar projections, something must be right with them this time.