Machine Learning Intern Neuromorphic & Edge AI

About

BrainChip is looking for a curious and technically‑driven Machine Learning Intern Neuromorphic & Edge AI. This is a unique opportunity to move beyond standard GPU‑heavy deep learning and dive into the world of Neuromorphic Computing. As an intern, you will help bridge the gap between traditional Artificial Neural Networks (ANNs) and BrainChip’s Akida™ event‑based processor. You’ll be working on the future of “AI at the Edge”—where power efficiency and real‑time learning are the ultimate goals.

Key Responsibilities

• Model Conversion & Benchmarking

  • Use BrainChip’s MetaTF™ framework to convert standard models (Keras/TensorFlow) into Spiking Neural Networks (SNNs).
  • Test and compare the accuracy and power consumption of models running on Akida hardware versus traditional CPUs/GPUs.
  • Help maintain and expand the Akida Model Zoo by training and validating new models for specific use cases (e.g., keyword spotting, gesture recognition).

• Data Pipeline & Quantization

  • Pre‑process datasets for event‑based processing (e.g., ImageNet, CIFAR, or specialized sensor data).
  • Assist in Quantization‑Aware Training (QAT) to ensure models maintain high performance at low bit‑widths (1, 2, or 4‑bit).
  • Experiment with “on‑chip” learning scenarios where the model adapts to new data without retraining in the cloud.

• Software & Tools Support

  • Write Python scripts to automate testing and performance profiling.
  • Document your findings and create tutorials or “Jupyter Notebook” examples to help our developers and customers understand neuromorphic workflows.

Qualifications & Skills

• Education: Currently pursuing a degree (B.S., M.S., or PhD) in Computer Science, Data Science, Electrical Engineering, or a related technical field.
• Core ML Knowledge: Solid understanding of Neural Network architectures (CNNs are a must; RNNs/Transformers are a plus).
• Programming: Proficient in Python. Familiarity with TensorFlow/Keras, PyTorch and ONYX is highly preferred.
• Mathematics: Comfortable with the linear algebra and calculus concepts behind backpropagation and optimization.
• The “Neuromorphic” Edge: You don’t need to be an expert in Spiking Neural Networks yet, but you should have a strong interest in biologically‑inspired AI and low‑power hardware.

Why Intern at BrainChip?

• Real Hardware Access: You won't just be running simulations; you’ll be deploying code onto physical Akida PCIe and SoC kits.
• Mentorship: Work directly with senior ML researchers and hardware architects who are pioneers in the neuromorphic space.
• Impact: Your benchmarks and model optimizations could end up in the hands of global customers building the next generation of smart devices.

Intern Program

• Duration: Approximately 3 months (flexible start depending on university’s summer schedule; e.g., May–June start, end Aug/Sept)
• Compensation: $28 per hour, 25 hrs per week