Purushothaman Saravanan
Senior Embedded Compiler Engineer at Latent AI
Princeton, New Jersey, United States
Summary
👤
Senior🎓
Top SchoolPurushothaman Saravanan is a Senior Embedded Compiler Engineer with eight years of experience focused on accelerating AI from training to on-device inference, currently developing real-time video segmentation and tracking systems for edge platforms like NVIDIA Jetson and Qualcomm Hexagon. He blends compiler and systems expertise (LLVM, CUDA, OpenMP) with hands-on model optimization using PyTorch, ONNX, TensorRT, TVM and quantization techniques (INT8/INT4/FP8), and has practical experience validating calibration and CI/CD for edge deployments. His background spans research in sparse tensor acceleration and parallel SpMM/SDDMM, industry internships in AI/ML and cloud, and a master’s in electrical engineering from NC State. Known for pragmatic experimentation—e.g., exploring AWQ INT4 on Jetson Thor and integrating quantization toolchains—he enjoys decentralizing training by leveraging device-level compute. Based in Princeton, he brings a rare combination of embedded compiler know-how and production-oriented AI deployment experience that shortens the path from model research to scalable edge inference.
8 years of coding experience
3 years of employment as a software developer
B.Tech, Electronics and Communications Engineering, B.Tech, Electronics and Communications Engineering at PES UniversityMaster of Science - MS, Electrical Engineering, Master of Science - MS, Electrical Engineering at North Carolina State University11th, 12th, PCME, 11th, 12th, PCME at Christ Junior college
Github Skills (14)
machine-learning9
deep-learning9
semantic-segmentation5
transcription5
midi4
shield4
controlling4
music4
pytorch3
gpio3
raspberry3
arduino2
raspberry-pi2
python2
Programming languages (2)
Github contributions (5)
Kodhandarama/CodeFundo_2019
Jul 2019 - Aug 2019
Contributions:69 commits in 1 month
Darth-Kronos/Spectrum-Sensing
Mar 2020 - Jan 2022
Using signal processing based features to train and validate machine-learning algorithms to improve spectrum sensing and related problems in cognitive radios.
Contributions:100 commits, 72 pushes, 3 branches in 1 year 10 months
cognitive-radiodeep-learningmachine-learningspectrum-sensing
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.