Full-Time
Backend Compiler Engineer New Grad
Posted on 8/27/2026
NVIDIA
10,001+ employees
Designs GPUs and AI HPC platforms
Compensation Overview
$108k - $195.5k/yr
Company Historically Provides H1B Sponsorship
Santa Clara, CA, USA
In Person
Bachelor's, Master's, PhD
 Software Engineering (1) |
|---|
Get referred to NVIDIA
See people who can refer or advise you
- A B.S., M.S., or Ph.D. degree, or equivalent experience.
- Compiler code generation experience, preferably with LLVM.
- Proficient hands-on C++ programming skills.
- A strong background in software engineering principles focused on crafting robust and maintainable solutions to challenging problems.
- Good communication and documentation skills.
- Self-motivation.
- Understand, modify, and improve the proprietary GPU compiler backend written in C++.
- Develop new register allocation compiler passes and optimizations to produce a robust, supportable compiler and tools.
- Work with global compiler, hardware, and application teams to oversee improvements and resolve problems.
- Contribute to deep-learning compiler technology spanning architecture design and support through functional languages.
- A significant academic compiler-related project or thesis.
- A background in LLVM code generation, including instruction scheduling, software pipelining, register allocation, GlobalISel, TableGen, LLVM intermediate representation, and Machine IR (MIR).
- Experience in compiler unit testing, intermediate representations, and visualization of intermediate representations.
NVIDIA designs and manufactures graphics processing units (GPUs) and computing platforms used for gaming, data centers, and artificial intelligence. These products work by using parallel processing to handle complex mathematical calculations much faster than standard computer processors, supported by a software ecosystem that allows developers to build and run AI models. Unlike competitors that may focus solely on hardware, NVIDIA integrates its chips with specialized software and cloud services to create a complete environment for high-performance tasks. The company’s goal is to provide the underlying technology necessary to power advanced computing, from realistic video game graphics to autonomous vehicles and large-scale data analysis.
Company Size
10,001+
Company Stage
IPO
Headquarters
Santa Clara, California
Founded
1993
Get referred to NVIDIA
See people who can refer or advise you
Simplify's Take
What believers are saying
- Q2 fiscal 2027 revenue hit $96.2 billion, up 106% year over year.
- Rubin production shipments start in 2026, targeting 10x lower inference costs.
- Anthropic's $10 billion NVIDIA stake and partnerships expand locked-in customer demand.
What critics are saying
- France's antitrust probe nears conclusion in July 2026, risking fines and conduct restrictions.
- U.S. export controls and China's port crackdowns threaten H20, H200, and Blackwell sales.
- OpenAI's $105 billion Ohio backstop and $500 billion financing draw circular-demand scrutiny.
What makes NVIDIA unique
- Blackwell and Rubin bundle GPUs, CPUs, networking, and software into one platform.
- NVIDIA shipped Vera Rubin samples in August 2026, staying ahead of AMD and Intel.
- Its data-center business reached $194 billion fiscal 2026, proving unmatched AI scale.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Company Equity
401(k) Company Match
Growth & Insights and Company News
6 month growth
↑ 0%1 year growth
↓ -1%2 year growth
↓ -2%An Oregon couple in their late 50s holds an Nvidia position now worth more than their home, 401(k)s, and all other assets combined. The stock has surged 864% over five years to around $221 per share. While Nvidia reported $96 billion in quarterly revenue with 75% gross margins, the concentrated position poses severe retirement risk. With a beta of 2.2, the stock has experienced double-digit drops following earnings despite strong results, including an 11% thirty-day decline after Q4 FY26. A 50% to 70% drawdown near retirement could force the couple to sell depressed shares for living expenses. Advisers recommend reducing concentration through staged selling across tax years, redirecting savings to index funds, or donating appreciated shares to donor-advised funds to avoid a single massive tax hit.
Nvidia has shifted its business model from primarily selling AI infrastructure to financing its customers' purchases. The company has invested nearly $50 billion in AI labs and arranged financing platforms targeting over $500 billion in outside capital. Management now describes Nvidia as a "full-stack AI factory platform" rather than the "data centre-scale AI infrastructure company" label used two years ago. The chip and networking segment still generates roughly $193.5 billion annually, about 64% of total revenue. The financing approach is clearest in Nvidia's NeoCloud arrangement, where it provides take-or-pay commitments on data centre capacity to help secure project funding, then shares rental revenue above a floor price. Days sales outstanding rose to 60 days in fiscal Q2 2027 due to extended payment terms for large customers. Inventory increased to $32 billion ahead of the Vera Rubin launch. Management forecast 70% revenue growth for fiscal 2028, constrained by supply.
Nvidia reported fiscal Q2 revenue of $96.2 billion, up 106% year-over-year, with data centre revenue rising 117% to $89.0 billion. CFO Colette Kress disclosed that cloud industry backlog now exceeds $2 trillion, with the top five hyperscalers expected to spend nearly $800 billion in 2026 and $1.3 trillion in 2027. Hyperscaler revenue reached $48.7 billion in Q2, more than double the prior year's $24.2 billion. At a $195 billion annual run rate, this represents approximately 24% of the projected $800 billion hyperscaler spending in 2026. The comparison carries caveats: Nvidia's fiscal year ends in late January, and its hyperscaler category includes more than five customers, suggesting the actual share runs somewhat lower.
Nvidia is rolling out DLSS 5, its AI-powered graphics technology, starting 3 September with NBA 2K27. The company calls it its largest computer-graphics advance since introducing real-time ray tracing in 2018. DLSS 5 uses what Nvidia terms 3D-Guided Neural Rendering. The neural model analyses scene objects to add realistic lighting, materials, shadows, and visual details. The technology's hardware requirements have dropped dramatically. When Nvidia demonstrated DLSS 5 in March, it required two flagship GeForce RTX 5090 graphics cards. Performance has improved fivefold, and the technology now runs on a single RTX 50-series GPU, including the RTX 5060. Nvidia states nearly every RTX 50-series GPU can run NBA 2K27 at 1080p with ray tracing and Ultra settings using DLSS 5.
Jim Cramer has called for NVIDIA to authorise a $500 billion share buyback programme, representing roughly 10% of the company, arguing that the stock does not reflect its earnings power. The suggestion comes as NVIDIA increasingly acts as a financier for AI infrastructure buildouts. NVIDIA reported fiscal Q2 2027 revenue of $96.2 billion, up 106% year-on-year, with data centre revenue reaching $89 billion. The company repurchased 203 million shares for $39.8 billion in the first half of fiscal 2027. Management expects approximately 70% revenue growth in fiscal 2028. However, NVIDIA faces mounting risks. Cramer noted that many companies NVIDIA backs are "non-investment grade". Reuters reported NVIDIA paused certain revenue-sharing agreements with AI cloud companies in August. The company also faces competition from Broadcom and custom chips developed by hyperscalers, alongside exposure to China's export restrictions.