Aditi Pandey
Software Engineer @Tenstorrent
Signup · Get unlimited contacts
WORK HISTORY
Software Engineer @Tenstorrent
Austin, TX, US
Working on ML compiler infrastructure (TT-MLIR, TTNN, StableHLO) for multi-device AI acceleration- Researching and implementing tensor sharding strategies across device meshes- Implementing and benchmarking collective communication ops including all_gather, all_reduce, and reduce_scatter- Comparing communication topologies (ring vs. linear) and analyzing compute vs. communication tradeoffs- Building reproducible MLIR-based workflows to auto-generate sharded graph variants- Profiling distributed execution and generating structured performance reports (CSV/analysis pipelines)- Contributing toward cost-model-driven distributed optimization, inspired by systems like Alpa, tailored for Tenstorrent hardware- Debugging and analyzing distributed graph transformations in Torch-XLA and StableHLO pipelines.
EDUCATION
University of Southern California
Master of Science - MS, Computer Science
MIT Academy of Engineering, Alandi, Pune
Bachelor of Technology - BTech, Computer Engineering
ABOUT ADITI PANDEY
As a Graduate Research Assistant at USC, my work focuses on the intersection of machine learning and high-performance computing. We\'ve achieved a 30% boost in model inference time through innovative ML Accelerator development and compiler optimizations. Previously, as an Associate Product Developer at BMC Software, I honed my skills in software development, data structures, and algorithms. My academic pursuits at the University of Southern California are paving the way for a future where technology seamlessly integrates with intelligent system design.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.