Adit Sheth

Amazon is offering substantial salaries, up to $263,700, to foreign hires for software engineering and other roles under O-1 and H-1B visas. https://lnkd.in/g4cyjH_2

5

1 Comment

Ruokai Yin, our summer intern from Yale to Microsoft Azure, worked with Xuan Zuo, @Sattwik Deb Mishra, Hokchhay T. and Preyas Shah and me. We published a paper based on his project: https://lnkd.in/eHuG3BSt Summary of paper: Finding optimal sharding is hard especially for large LLMs. A large number of devices have to be divided into TP, PP, EP dimensions. For TP, we have to decide the tensor dimension e.g. head, hidden size to shard for every op. Additionally, TP can be one-dimensional/two-dimensional etc. This makes for a large sharding space that is hard to cover exhaustively. SoTA methods use heuristics to select sharding e.g. GSPMD, Sequence Parallelism, Helix etc. There are significant time gaps, months or years, between the discovery of each of these methods. We want to fast-forward the process of discovery by 1) using a formal representation of the sharding space, and, 2) using reinforcement learning to search the sharding space.    This paper demonstrates that our method is 3.5x better than random search, and on par with hand-tuned heuristics when looking only at 1D sharding and without allowing sequence parallelism. We aim to extend the work to multi-dimensional sharding as well as sequence parallelism. Beyond that, we wish to combine the problem of sharding with fusion, tiling and collectives as well.

58

The H1B fee hike didn’t just raise the cost of talent. It exposed the fragility of relying on immigration timelines to scale core engineering teams. That approach was never sustainable. Now it’s also impractical. We’ve shifted. AI engineers are embedded directly into overseas teams. No layers. No lag. Same expectations on code quality, sprint velocity, and reviews—just without the bottleneck. Teams stay lean, fast, and accountable. This isn’t about offshoring. It’s about fixing a broken system. If your team model still depends on borders, it’s already out of date. #AIEngineering #TechLeadership #RemoteTeams #PostH1B #ScalableTalent #BuildEfficiently #DistributedEngineering #ContinuousImprovement #FutureOfWork

4

President Trump has signed an executive order imposing a $100,000 annual fee on H-1B visas, dramatically reshaping U.S. immigration and labor policy. Since 70% of H-1B visa holders are Indians, the impact will be disproportionately felt by Indian professionals and companies with U.S. operations. The administration argues the move protects American workers and raises the cost of hiring foreign talent, ensuring only the “best and brightest” qualify. However, the implications are far-reaching. U.S. employers — especially in technology, consulting, and research — will see labor costs surge. For large corporations the fee may be manageable, but startups and mid-sized firms could be priced out of access to specialized global talent, pushing them to offshore projects instead. Indian workers already in the U.S. face uncertainty, while those abroad may be unable to enter unless employers can justify costs or secure exemptions. The policy risks slowing innovation, weakening America’s competitive edge, and straining U.S.–India talent pipelines. Critics question whether the executive branch can legally impose such a steep fee without Congressional approval. Lawsuits are expected. For now, the order signals a more protectionist stance and could redirect global skilled migration toward countries like Canada, the UK, and Australia.

18