Why This Matters Now

In January 2024, Zuckerberg publicly committed to developing what he referred to as “general intelligence,” intending to share it more broadly than his competitors. He highlighted tools like Llama as central to this strategy, supported by a substantial boost in computing power and research investment. Unlike other labs that take a more closed approach, Meta is focused on assembling a distinguished team to create more capable AI systems.

“We are committing to building general intelligence… and open sourcing it responsibly so everyone can benefit.”

– Mark Zuckerberg, January 2024

Since then, Meta has ramped up its investment in AI infrastructure, consistently launching larger and more capable open models, while quietly recruiting senior researchers and engineers to enhance capabilities and safety.

What Is Meta’s “Superintelligence” Team?

“Superintelligence” isn’t an official product title; rather, it serves as shorthand for Meta’s internal initiative aimed at advancing beyond current large language models toward more general, multimodal systems. This group integrates research, engineering, infrastructure, and safety to scale model capabilities and reliability across Meta’s products and developer ecosystem.

WIRED has been monitoring notable hires as this effort progresses, compiling an ongoing list of new recruits and internal shifts. While these rosters can change rapidly, their reporting provides a valuable snapshot of how Meta is staffing this initiative and where the talent is sourced from. For the latest updates, check out WIRED’s live coverage linked below.

The Leaders Shaping Meta’s AI Push

While Meta hasn’t disclosed an official organizational chart for the “superintelligence” team, several leaders are publicly guiding key components of the strategy:

• Mark Zuckerberg – CEO. He has directed efforts to build general intelligence and open-source components safely, while also increasing compute and research funding [The Verge].
• Yann LeCun – Chief AI Scientist. He leads long-term AI research across Meta, including work on architectures that go beyond current LLMs, promoting learning-based systems capable of reasoning and planning, and advocates for open science in AI.
• Joelle Pineau – VP, AI Research. She oversees Meta AI Research (FAIR) worldwide, covering language, vision, multimodal, and responsible AI research.
• Ahmad Al-Dahle – VP, Generative AI. He manages generative AI product and platform teams, connecting research breakthroughs to products like Meta AI, creation tools, and offerings for developers.
• Llama Model Leadership – Researchers and engineers who have delivered Llama 2 and Llama 3, including notable contributors like Hugo Touvron, are driving the development of state-of-the-art open models [Meta AI].
• Infrastructure and Systems – Meta’s infrastructure organization has expanded its AI clusters, networking, and custom silicon programs to support unprecedented scales of training and inference [Reuters].

These leaders and their teams are actively hiring across research, engineering, and safety to advance the next wave of capabilities and enhance systems for real-world applications.

The Types of Experts Meta Is Hiring

From public job postings, research releases, and industry news, here’s a snapshot of the profiles Meta is targeting for its superintelligence initiative:

• Foundation Model Researchers – Experts in pretraining, scaling laws, optimization, and data curation for text, vision, audio, and multimodal models.
• Reasoning and Tool-Use Engineers – Specialists in agents, retrieval-augmented generation, program-of-thought, and external tool integration to enhance reliability and task completion.
• Safety and Alignment Researchers – Professionals focused on red-teaming, evaluations, interpretability, and scalable oversight to minimize harmful outputs and align models with ethical guidelines.
• Efficiency and Systems Experts – Engineers who drive distributed training, compiler stacks, inference optimization, and quantization to make large models viable at Meta’s scale.
• Applied Product Builders – Teams that incorporate models into Meta AI, creator tools, messaging, and enterprise workflows, focusing on latency, privacy, and safeguard measures.

Industry reports indicate that Meta is offering highly competitive packages to attract senior AI talent from leading labs, highlighting the fierce competition for these skills [The Information].

What We Can Say Confidently About the Hires

Although the arrivals of new team members change frequently, several trends are emerging:

• Cross-Pollination from Leading Labs – Meta is attracting researchers and engineers with backgrounds from OpenAI, Google DeepMind, Anthropic, and top academic groups, which brings new perspectives and accelerates knowledge transfer.
• Heavy Investment in Infrastructure Hires – As Meta scales to train and support larger multimodal models, it is focusing on improving distributed training, data pipelines, and GPU efficiency. The company has raised its 2024 capital expenditure guidance to the mid-high 30s billions, chiefly due to AI infrastructure [Reuters].
• Open-Model Strategy Remains Central – The release of Llama 3 has reaffirmed Meta’s commitment to open models, which in turn attracts contributors who prioritize open science and broader developer impact [Meta AI].
• Safety Measures Are Becoming More Formalized – Meta has increased its focus on red-teaming, evaluations, and policy enforcement for its models and assistant features, and is actively hiring to expand these functions alongside capability teams.

How Meta’s Approach Compares

When compared to other AI powerhouses, Meta’s superintelligence initiative reveals two key differentiators:

• Open-Source Orientation – Meta’s open releases of the Llama family stand in contrast to the more closed access maintained by OpenAI and Google. This approach can hasten external research and adoption, but it also necessitates careful safety practices and licensing.
• Consumer-Scale Integration – Since Meta’s models power tools and assistants across Facebook, Instagram, WhatsApp, and Quest, the hiring focuses on performance at scale, low-latency inference, and policy enforcement across billions of users.

Zuckerberg’s public framing of “general intelligence” and the computational resources to support it sets expectations that Meta will continue to enhance model size, data quality, and multimodality. He has mentioned plans to build one of the industry’s largest GPU fleets to support this direction [CNBC].

What This Means for Developers and Professionals

If you are involved in building or evaluating AI systems, here’s how Meta’s recruitment and strategy may impact you:

• Faster Updates of Open Models – Anticipate further Llama updates and new multimodal checkpoints that widen your capabilities without being restricted by closed APIs.
• Enhanced Reliability and Reasoning – An increase in hires focused on safety, evaluations, and tool-use should lead to steady improvements in accuracy, refusal behavior, and task fulfillment.
• Better Performance for Your Budget – Systems and efficiency engineers will boost inference throughput, making on-premises and edge deployments more practical.
• Clearer Guidance and Safeguards – Expect more comprehensive documentation, usage guidelines, and model cards as Meta formalizes its safety processes.

Bottom Line

Meta’s superintelligence team is less of a single entity and more of a coordinated effort across research, infrastructure, and products aimed at making general-purpose AI more powerful and widely available. This recruiting surge, along with increasing AI capital expenditures and consistent releases of open models, indicates that Meta plans to be among the few organizations developing frontier-scale systems. If you’re tracking the AI talent landscape or planning to build with open models, this is definitely a story to keep an eye on.

FAQs

Is Meta really building AGI or superintelligence?

Meta uses the term “general intelligence” rather than AGI. This essentially means they are working on models that can reason across different tasks and modalities with fewer transitions, making them accessible to developers and users alike [The Verge].

Is there an official roster of Meta’s superintelligence team?

No official roster has been released. WIRED is keeping an ongoing list of hires and internal movements as they are publicly confirmed or reported. Check their coverage linked below.

How does Meta’s approach differ from OpenAI or Google?

Meta places a strong emphasis on open models and tools, whereas OpenAI and Google generally keep their most powerful systems behind APIs. Additionally, Meta integrates models across its consumer applications at a global scale, influencing its hiring focus.

Will Meta open source future, more capable models?

Zuckerberg has stated that Meta will open source responsibly, which suggests that decisions will vary based on capability, misuse risks, and safety measures. The releases of Llama imply a continuing commitment to openness with appropriate safeguards.

What kinds of roles is Meta hiring for?

Roles include foundation model research, applied reasoning and agents, safety and alignment, efficiency and systems, as well as product engineering across Meta AI and creative tools. Public postings and research updates cover all these areas.

Sources

1. WIRED – Here Is Everyone Mark Zuckerberg Has Hired So Far for Meta’s Superintelligence Team
2. The Verge – Mark Zuckerberg says Meta is building AGI and will share it
3. CNBC – Zuckerberg says Meta is building general intelligence, details compute plans
4. Meta AI – Introducing Llama 3
5. Reuters – Meta significantly ramps up AI spending
6. The Information – Meta courts AI researchers with million-dollar pay packages