Research Intern - Neural Network Training and Hardware–Software Co-Design
Meta PlatformsJob Title
Research Intern - Neural Network Training and Hardware–Software Co-Design
Role Summary
Internship on a research team focused on designing and training neural networks optimized for custom hardware used in wearables and robotics. The intern will co-design models and hardware-aware training flows, prototype on FPGA platforms, and validate performance against tight accuracy, latency, power, and memory targets.
Experience Level
Internship / Entry-level role for current degree-seeking students. Specific academic eligibility details are in the Education Requirements section.
Responsibilities
Work end-to-end on model and system development for constrained hardware platforms.
- Design, train, and evaluate neural networks for on-device workloads under compute, memory, and energy constraints.
- Develop hardware-aware training and model compression techniques and measure impact on target hardware.
- Explore partitioning models across multiple accelerators or chips and evaluate inter-chip communication trade-offs.
- Profile models end-to-end and use measurements to guide model and hardware decisions.
- Prototype and validate on FPGA platforms; modify and debug existing RTL as needed.
- Present results via a working demo and a technical write-up.
Requirements
Core qualifications and logistical requirements for the internship.
Must-have:
- Hands-on experience training deep neural networks using PyTorch or TensorFlow, including data pipelines, debugging, and evaluation.
- Working knowledge of computer architecture, including how memory hierarchy, data movement, and numerical precision affect neural network performance.
- Ability to prototype and validate on FPGA platforms and debug RTL as needed (practical FPGA/RTL experience preferred but not strictly required).
- Must obtain and maintain work authorization in the country of employment for the duration of the internship.
Nice-to-have:
- RTL design experience in Verilog or SystemVerilog, including simulation and testbenches.
- FPGA experience: synthesis, timing closure, on-board bring-up and debug.
- Experience with quantization-aware training, mixed precision, pruning, or knowledge distillation.
- Familiarity with neural network accelerators, model parallelism, distributed inference, and embedded/edge deployment (C/C++).
- Publications or substantial projects in efficient ML, ML systems, or computer architecture (e.g., MLSys, NeurIPS, ICLR, ISCA, MICRO, DAC).
Education Requirements
Currently pursuing a PhD in Computer Science, Electrical/Computer Engineering, Machine Learning, or a related field; intent to return to a full-time degree program after the internship. (No alternative-equivalent-experience language provided.)
About the Company
Company: Meta Platforms
Headquarters: Menlo Park, California, United States
American technology company that develops social networking products (Facebook, Instagram, WhatsApp) and invests in virtual/augmented reality hardware and software through Reality Labs, focusing on connectivity, advertising, and immersive computing experiences.
