ML Engineering Intern

Matrice AI; United States; Remote; USD 150,000 - 190,000; Full time

Job summary

Matrice AI is seeking a highly skilled ML Engineer to design and optimize training and inference systems for cutting-edge vision-language and multi-modal models. You will implement neural architectures from first principles and ensure efficiency on edge and cloud platforms.

You’ll own end-to-end model implementations, deepen your math understanding, and trade off performance with memory and speed. Expect a hands-on, high-ownership role in a fast-moving field.

Qualifications

• Expert-level proficiency in Python across problem domains.
• Strong CS fundamentals: memory management, multi-threading, distributed systems.
• Extreme ownership and accountability for code quality and system performance.
• Willingness to learn the mathematics behind models, not just use libraries.

Responsibilities

• Design training and inference pipelines for state-of-the-art models.
• Write performant inference code bridging research and production.
• Implement neural architectures from first principles using Python/NumPy.
• Debug and optimize for speed, memory, and accuracy.

Skills

Python

Multi‑threading

Memory management

Distributed systems

Code quality

Ownership

Tools

NumPy

PyTorch

Job description

Batch: 2026/2027/2028.

About the role:

• Build training and inference systems that push the boundaries of modern machine learning
• Engage with cutting-edge architectures spanning Vision Language Models, multi-modal systems, and generative AI technologies
• Take ownership of model implementations by mastering the underlying mathematics rather than relying on framework abstractions alone
• Optimize models to run efficiently across both edge devices and cloud infrastructure
• Ground your work in fundamental principles rather than treating libraries as black boxes

What you'll work on:

• Designing and refining data pipelines that feed state-of-the‑art models through their training cycles
• Writing performant inference code that bridges research and production
• Implementing complex neural network architectures from mathematical first principles
• Debugging and optimizing systems for speed, memory usage, and accuracy

What we're looking for:

• Expert-level proficiency in Python across all problem domains
• Deep understanding of computer science fundamentals including multi-threading, memory management, and distributed systems concepts
• A mindset of extreme ownership and accountability for code quality and system performance
• Willingness to learn the mathematics behind your models rather than treat them as utilities

Selection process:

• Coding challenge requiring you to solve a complex algorithmic problem (Medium/Hard LeetCode level) in Python
• Technical interview where you'll implement core ML components from scratch using only Python and NumPy—such as building a Transformer block or Convolution layer—to demonstrate your grasp of both math and engineering