Tier-3 ISE final year(2026 batch) with ongoing ML research (EMSE/Q1/NeurIPS/A* target), trying to understand real impact in India [D]

I went through a bunch of older posts here about research vs dev roles, but most of them were either very general or not really in a similar situation, so posting this. I’m a final year ISE student from a tier-3 college. Over the past 1.5–2 years I’ve been focusing quite a bit on ML research instead of just the usual DSA + dev route. Current situation: 1 paper in EMSE 1 in JAIR 1 under production still but hopefully going to a Q1 journal 1 I’m trying for NeurIPS main track (I know this one’s a long shot) -> Already under review and didn't get bench rejected 2 month internship at Accenture in 3rd year Some ML projects apart from the research work I know not everything will land. But assuming a realistic outcome where maybe 1–2 of these get accepted at a decent level (Q1/A* types), I’m trying to figure out what that actually changes. A few things I’m confused about: For jobs in India: Does this actually help with shortlisting for ML/SDE roles, or after a point does it not matter much and it just comes down to DSA + interviews anyway? Also, being from a tier-3 college, does this help offset that at all? Or do companies still filter heavily based on college first ? I've seen a lot of people from IIT/NIT get into research/MLE roles in big MNC's rather than preferring those in Tier-3 colleges. For higher studies: Does having papers like this make a noticeable difference for PhD abroad (US/EU), or is it just a “nice to have”? Do colleges really care about the difference between something like NeurIPS vs a Q1 journal vs IEEE Access, or is it all seen more or less similarly? And finally, I'm planning to do my M.tech in India itself by writing GATE 2027, do you recon the value of these paper will actually help me in these colleges (say IIT's/IISC/NIT ?) And one thing I’m seriously unsure about: If I’m leaning towards industry (ML/AI roles), is continuing research actually worth the time, or would that effort be better spent on DSA, systems, etc? Also, is it even realistic to