Ambient Scientific
Website:
ambientscientific.ai
Job details:
Key Responsibilities:
•
Lead design and development of the AI/ML compiler toolchain for Ambient’s ultra-low power AI chips.
•
Build and optimize compiler backends, IR transformations, and runtime systems for AI workloads.
•
Drive model conversion pipelines (TF, TFLite, PyTorch, ONNX → Ambient compiler stack).
•
Implement graph-level and operator-level optimizations (fusion, scheduling, memory reuse).
•
Collaborate with hardware and data science teams to co-design hardware-aware compiler strategies.
•
Develop tools for quantization, pruning, mixed precision, and efficient memory allocation.
•
Ensure compiler generates binaries with predictable latency, power, and memory footprint.
•
Define performance metrics and benchmarking strategies for compiler outputs.
•
Mentor a team of compiler and software engineers; review design, code, and optimization flows.
•
Stay up to date with MLIR, LLVM, XLA and latest AI compiler trends.
Key Requirements:
•
Bachelor’s/Master’s in CSE, ECE, EEE, or related.
•
7+ years of experience in compiler development, with at least 3+ years in ML compilers/AI acceleration.
•
Strong expertise in compiler internals: parsing, IR design, optimization passes, code generation.
•
Proficiency in C/C++, Python, LLVM/MLIR, ONNX Runtime.
•
Solid understanding of neural network architectures, graph optimization, and hardware acceleration.
•
Familiarity with DSPs, GPUs, and AI accelerators; knowledge of edge hardware (ARM Cortex-M, RISC-V) desirable.
•
Experience in runtime systems, scheduling, and memory management for constrained devices.
•
Excellent leadership, problem-solving, and cross-functional collaboration skills.
Click on Apply to know more.
