Aiswarya Sankar — Founder and CEO @ Entelligence.AI (We’re Hiring!) | Ex-Uber
Founder and CEO @ Entelligence.AI (We’re Hiring!) | Ex-Uber
Aiswarya Sankar ranks #746 of 19,530 LinkedIn creators in Computer Software, and is a standout voice in United States. They have 18.7K followers and published 12 posts in the last 30 days at a 0.5% average engagement rate.
- 18.7K followers
- 12 posts / 30d
- 0.5% avg engagement
- 577 follower growth / 30d
The roast
Aiswarya founded Entelligence.AI to automate engineering, which explains why her content strategy is currently indistinguishable from a dead server. With 18,000 followers and less engagement than a local librarian’s knitting circle, she’s the only AI founder who successfully automated her own relevance.
About Aiswarya
Building artificial engineering intelligence at entelligence.ai. Join us!
Highlights
- Top 5% in Computer Software — Ranked #158 of 4763 creators
- Top 5% in United States — Ranked #297 of 6030 creators
- Top 10% Audience — 18,689 followers
- Top 10% Creator — 14 posts in 30 days
Recent posts
Some thoughts after spending the last 3 days talking to engineering leaders at the AI Engineering World's Fair - - Leaders currently don't have the tools to measure AI eng ROI on a project, outcome basis - Token maxxing is dying very quickly at an enterprise level - Claude code isn't a one shot solution for code gen and review - Bills are growing exponentially - the average 100 person team spends $500k + annually - Current AI budgeting is extremely haphazard and there are no good ways to budget and allocate per teams - Eng leaders are focused on standardizing skill generation across teams
78 reactions · 3 comments · 3 reposts
Measuring ROI from increased token spend isn't just counting PRs. It's actually attributing sessions to projects and outcomes - something every eng leader we speak to has been repeatedly telling us is completely opaque with current tools today. In this conversation, our founding engineer Adithya Saravu and I break down what we're seeing across 2,000+ companies and how eng leaders will need to think about token budget allocation going forward. The fix isn't cutting or capping usage - it's allocating budget the way you'd allocate headcount: per project, per priority, with real predictive guard
83 reactions · 3 comments · 5 reposts
Interesting results from a head-to-head my team ran: GLM-5.2 vs Claude Opus on 45 terminal bench tasks, same agent, same harness, only the model swapped. GLM-5.2 solved 25/45, the same number of tasks as Opus, burned ~3.3x more tokens, but landed at only 46% of the cost thanks to far cheaper per-token pricing. Open weights have reached the frontier on real coding-agent work. https://lnkd.in/gENhJ-6w
39 reactions · 1 comments · 2 reposts
Last updated 2026-08-01