Role/Responsibilities:
EnCharge is looking for a senior software engineer (and technical lead) who can help us build out Tensor Virtual Machine (TVM) based end-to-end software runtimes for the next generation of Edge AI Hardware. You must have a track record of building runtimes & compiler stacks for AI Inference hardware, experience with TVM, as well as strong cross-functional contributions to AI architectures & software stacks.
- Enabling TVM support for EnCharge AI hardware, compiler, neural network operations (ops), compiler optimizations and runtime APIs.
- Working closely with the compiler teams to define the APIs and graph exchange formats needed for TVM to connect with the EnCharge Compiler stack (to build optimized binaries).
- Optimizing end-to-end performance of the TVM runtime on EnCharge hardware platforms.
- Enabling TVM backends for MLPerf benchmarking.
- Mentor / lead junior engineers across the company.
Qualifications/Required Skills:
- Masters/Ph.D. in EE/CS with >5 years of experience in AI applications, compilers & hardware.
- Proficiency in C++ & Python.
- Deep experience with modifying the TVM code base to support new hardware backends.
- At least 3-5 years of experience with Tensorflow, PyTorch.
- >2 years of experience with Deep Learning compilers and strong experience with AI compiler optimizations.
- Solid understanding with state-of-the-art neural network topologies in various application domains (and especially in the computer vision space).
- Excellent verbal and communication skills.
Preferred/Beneficial Skills:
- Knowledge of industry-standard (and advanced) tools, graph, and intermediate-representation (IR) formats and methodologies including LLVM, MLIR etc.
- Open-source experience.