About the role#
We are looking for a Machine Learning Intern to join our Software - ML & Algorithm team. You will work on training, compressing, and deploying neural networks on our analog compute-in-memory chips. You will take models from PyTorch checkpoints to versions that run on our hardware, while measuring accuracy and efficiency trade-offs. You will work alongside our model, compiler, and hardware teams.
What you'll need#
- Currently pursuing a BS, MS, or PhD in Computer Science, Electrical Engineering, or a related field.
- Hands-on experience training and evaluating neural networks using PyTorch through research, internships, or substantial projects.
- Familiarity with at least one model optimization technique like quantization, pruning, distillation, or efficient architecture design.
- Solid understanding of deep learning fundamentals, including neural network layers, normalization, attention mechanisms, backpropagation, and numerical precision.
- Strong Python skills and familiarity with software development practices like version control, debugging, and testing.
- Ability to read technical research papers, implement published methods, and evaluate their applicability.
- Experience with edge or embedded targets, ML compilers, C++, or audio/speech and compact vision models is a plus.
What you'll do#
- Assist in training, fine-tuning, and adapting neural network models for compute-in-memory hardware.
- Implement and evaluate model compression techniques, including post-training quantization, quantization-aware training, mixed-precision assignment, pruning, and knowledge distillation.
- Develop hardware-aware training methods to make models robust to low-precision weights and analog non-idealities.
- Analyze model graphs for operator coverage, rewrite unsupported operators, and export models for the compiler toolchain.
- Profile accuracy, latency, and memory on simulation and hardware to identify where accuracy is lost.
- Write clean, tested Python and C++ for model training, quantization, and evaluation tools.
- Document experimental methodologies and present technical findings to the team.
Location & details#
- Location: San Jose, California, United States.
- Modality: On-site.
- Compensation: $34 - $45 per hour.
- Employment type: Full-time.
- Term: Rolling.


