Full-Time
Member of Technical Staff
Kernels
Magic
51-200 employees
AI-powered code generation and collaboration
Compensation Overview
$225k - $550k/yr
H1B Sponsorship Available
Remote in USA + 1 more
More locations: San Francisco, CA, USA
In Person
 Software Engineering (1) |
|---|
- Low-level programming experience targeting AI accelerators such as NVIDIA Blackwell or Google Tensor Processing Units
- Develop and optimize GPU kernels in frameworks such as NCCL, MSCCLPP, CUTLASS, CuTeDSL, Triton, Quack, Flash-Attention, and similar frameworks
- Experience in other kernel authoring frameworks such as Pallas/Mosaic (GPU) or Pallas/Mosaic (TPU), or Mojo also maps well to the work on Magic's kernel team
- Strong depth over shallow breadth: for kernel engineering, we prefer candidates with deep expertise in computer architecture, low-level machine optimizations, and code generation, with breadth across machine learning
- Agility, ownership mindset, and grit
- Design, implement, and maintain high-performance kernels to optimize throughput and latency during training and inference
- Design and implement kernels that support high-performance long-context behavior
- Ownership of kernel design, implementation, deployment, and production reliability
- Focus on robustness, extensive testing, and functional correctness, while pushing on performance
- Evaluate porting Magic’s compute kernels to alternative hardware options
- Co-design kernels with understanding and interaction with training, inference, and RL teams
Magic.dev focuses on building AI-powered coding tools that act as a virtual coworker for developers. It creates proprietary models that understand and generate code, enabling collaborative coding tasks rather than simple assistance. The product works through subscription or licensing, allowing clients to integrate the AI into their own systems and workflows after training on large datasets with extensive R&D. What sets Magic.dev apart is its emphasis on collaboration with developers as a true coding partner, backed by frontier-scale models, heavy data collection, and a scalable infrastructure designed for enterprise needs. The company’s goal is to advance AI in software development by enabling its virtual coworker to generate, optimize, and co-create code with human programmers.
Company Size
51-200
Company Stage
Late Stage VC
Total Funding
$465M
Headquarters
San Francisco, California
Founded
2022
Simplify's Take
What believers are saying
- Raised $320M August 2024 from Eric Schmidt, Sequoia, enabling scaling to 160 exaflops compute capacity.
- Enterprise customers like Atlassian and Canva validate product-market fit and provide distribution leverage.
- Frontier models with domain-specific reinforcement learning and inference-time compute improve code generation accuracy over time.
What critics are saying
- Claude and o1 reasoning models capture enterprise clients seeking integrated AI assistants, not standalone vendors.
- GitHub Copilot's Microsoft ecosystem integration locks developer workflows; Magic lacks equivalent distribution channels.
- $515M burn rate requires sustained revenue growth; insolvency risk within 18–24 months if acquisition slows.
- Open-source models like Llama matching Magic's performance collapses pricing power and forces commoditization.
- Key researcher departures to well-capitalized competitors cripple R&D velocity and model development.
- EU AI Act compliance and copyright litigation over training data force costly model retraining and delays.
What makes Magic unique
- Ultra-long context windows via proprietary Long-term Memory Network architecture differentiate Magic from GitHub Copilot and competitors.
- Frontier-scale infrastructure partnership with Google Cloud and Nvidia Blackwell enables fastest timeline to scale compute.
- Focus on safe AGI through automating AI research and code generation attracts mission-aligned capital and talent.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Health Insurance
Dental Insurance
Vision Insurance
Unlimited Paid Time Off
401(k) Retirement Plan
401(k) Company Match
Company Equity
Relocation Assistance
Growth & Insights and Company News
6 month growth
↑ 2%1 year growth
↑ 1%2 year growth
↑ 0%Australia's largest private tech company, Canva, has made its ninth acquisition by purchasing the AI marketing group Magic. This move comes as Canva reports a rise in paying users and approaches its IPO.
Startups that build artificial intelligence (AI) applications on top of large language models (LLMs) are reportedly making rapid gains in both sales and funding. These startups are reaching as much as $200 million in annual recurring revenue in less than two years and, as a group, increased the amount of funding they attracted by 110% to reach $8.2 billion in 2024, the Financial Times (FT) reported Monday (April 14), citing data from Dealroom.co and Flashpoint. This category of AI startups includes companies like Perplexity, Synthesia, ElevenLabs, Harvey and Sierra, according to the report
Flat invest in people who are passionate, dedicated, have potential and skin in the game. Founders that care deeply about their customers and their businesses.
In 2024, European AI companies raised nearly €3 billion across 137 deals, a 35% increase from 2023. French companies led with over €1.3 billion, followed by Germany with €910 million, and the UK with €318 million. The top deal was Mistral AI in France, securing €600 million. Other notable deals included Helsing in Germany with €450 million and Poolside in France with $400 million. Gcore in Luxembourg raised $60 million, ranking tenth among the biggest deals.
Flat Capital has invested SEK 9 million in Magic AI, an AI and code generation company, by purchasing existing shares. Magic, founded in 2022 by Eric Steinberger, aims to achieve secure Artificial General Intelligence. The company has raised $515 million (SEK 6 billion) so far, with backing from investors like Sequoia Capital and Eric Schmidt. Hanna Andreen, CEO of Flat Capital, expressed excitement about the potential industry impact of Magic's innovations.