Full-Time

GenAI Performance Engineer

Posted on 1/13/2026

Modular

Modular

201-500 employees

Unified AI infrastructure platform for workloads

Compensation Overview

$167k - $286k/yr

+ Bonus + Equity

Remote in Canada + 1 more

More locations: Mountain View, CA, USA

Remote

Onboarding conducted in-person at Los Altos, CA; occasional travel 2-4 times per year.

Category
AI & Machine Learning (1)
Required Skills
Python
Requirements
  • 5+ years of professional or postgraduate academic experience working on or researching performance analysis, tooling or benchmarking
  • Expertise in performance measurement (i.e. benchmarking), modeling, and analysis on real-world workloads
  • Extensive experience with Python
  • Creativity and curiosity for solving complex problems
  • Experience writing production-quality software
  • Strong written and verbal communication skills
Responsibilities
  • Measure, analyze, and identify opportunities to improve the performance of the MAX product under realistic and relevant usage patterns
  • Partner with the product and customer teams to understand the performance of the MAX product in both standard and cutting edge AI applications and design benchmarks to reflect them
  • Collaborate with the kernels and GenAI modeling team to bring up new model families
  • Collaborate with the kernels and runtime team to bring up and optimize new GPUs and accelerators
  • Collaborate with the cloud team to design and benchmark advanced serving features and new serving algorithms
  • Build statistical models and tools to operate on benchmarking and telemetry data and help develop key insights for performance, cloud costs, etc.

Modular builds and offers a unified AI infrastructure platform that teams can use to develop, deploy, and innovate on AI workloads. The platform is designed to be integrated and extensible, providing a suite of tools that work together to simplify infrastructure and accelerate AI work. It charges clients for access to the platform and its tools, focusing on businesses with AI infrastructure needs in technology, engineering, and finance. A key differentiator is direct access to industry experts and the platform’s support for dynamic shapes in AI workloads, which helps meet SLAs and SLOs. The company aims to help teams move faster by making AI infrastructure easier to manage and scale, while fostering a culture centered on building user-loved products, empowering people, and teamwork.

Company Size

201-500

Company Stage

Late Stage VC

Total Funding

$380M

Headquarters

Palo Alto, California

Founded

2022

Simplify Jobs

Simplify's Take

What believers are saying

  • Raised $380M in September 2025 at $1.6B valuation from Greylock, GV.
  • Partners with Inworld AI for speech synthesis and NVIDIA for CUDA integration.
  • Open-source Modular Platform installs via pip, supports Llama models instantly.

What critics are saying

  • Nvidia absorbs cross-platform tech into CUDA within 12-24 months.
  • AMD ROCm and Intel oneAPI achieve parity, eroding Modular's edge in 18-36 months.
  • Hyperscalers like Google, Meta build proprietary stacks, blocking 60% AI spend.

What makes Modular unique

  • Mojo programming language enables Python-like usability across Nvidia, AMD GPUs.
  • Achieves top performance on Nvidia Blackwell B200 and AMD MI355X seamlessly.
  • Unified MAX Platform optimizes from GPU kernels to cloud APIs hardware-agnostically.

Help us improve and share your feedback! Did you find this helpful?

Benefits

Health Insurance

401(k) Company Match

Unlimited Paid Time Off

Stock Options

Growth & Insights and Company News

Headcount

6 month growth

-1%

1 year growth

-2%

2 year growth

0%
Business Insider Japan
Dec 22nd, 2025
Modular raises $380M from top VCs to challenge Nvidia's AI dominance with new software stack

Modular, a startup co-founded by Apple and Google software veterans Chris Lattner and Tim Davis, has raised $380 million from investors including Greylock, General Catalyst and GV, reaching a valuation of $1.6 billion in its latest September funding round. The company is challenging Nvidia's CUDA software platform, which has dominated AI development for nearly 20 years by binding workloads to Nvidia GPUs. Lattner, known for creating Apple's Swift programming language, and Davis previously built software for Google's TPU AI chips. Modular has developed a new AI software stack including Mojo, a programming language designed to work across different chip manufacturers whilst maintaining Python-like usability. The company recently announced achieving top-level performance on both Nvidia's Blackwell B200 and AMD's MI355X GPUs using the same software platform, with AMD chips showing approximately 50% better performance than when using AMD's own software.

Modular
Nov 27th, 2025
Modular: Modular Raises $250M to scale AI's Unified Compute Layer

Modular Raises $250M in Third Round to Unify AI Compute

PYMNTS
Sep 29th, 2025
Big Checks Flow to AI's Hidden Foundations as Investors Look Beyond Models

Backers in Modular's third funding round included U.S. Innovative Technology fund, DFJ Growth, Google Ventures, General Catalyst and Greylock Ventures, the release said.

SiliconANGLE Media
Sep 24th, 2025
Modular raises $250M to simplify AI deployment across hardware

Modular has also teamed up with AI application developers such as Inworld AI to accelerate speech synthesis and San Francisco Compute Co., which operates a GPU cluster marketplace.

TechStartups.com
Apr 7th, 2025
From .Ai To .Com: The Quiet Domain Rebrand Sweeping Startup Ecosystem

When generative AI took off in 2022 following the popularity of ChatGPT, launching a startup on a .ai domain felt like the obvious move. It signaled your company was part of the next wave of innovation.¬†It instantly signaled to investors, journalists, and users that you were building in AI.Now, a shift is happening. Some of the most promising startups in AI are quietly moving away from their .ai domains and switching to .com. It‚Äôs not about trend-chasing‚Äîit‚Äôs about positioning.This isn’t a flood, but it‚Äôs enough to raise eyebrows. And it says something about how branding, trust, and long-term ambition play a bigger role in how startups choose to present themselves.Why Startups Went With .AI DomainsIn the early days, .ai domains were easy to get and made a lot of sense. Short .coms are hard to come by and often expensive

INACTIVE