Planning a large training run still takes deep expertise. Someone has to translate a goal, such as a model size, a dataset, and a deadline, into GPU counts, parallelism settings, memory, storage, and a cluster to run on. Learning a new operations platform adds another hurdle, and every week spent learning it is a week the AI data center is not running at its best.
Federator.ai Cortex removes both barriers. The Intent Compiler turns a one-line description of a workload into a costed, ready-to-submit resource plan, while the Megatron-LM integration gives engineers full control when they want it. Wingman AI, a copilot built into Federator.ai Cortex, answers questions about the platform and the fleet in plain language, so new teams become productive in days rather than months.
Intent Compiler: From Intent to Resource Plan
Describe the workload the way you would explain it to a colleague, for example “train a 7B model on 8 H100 GPUs for 3 days.” The Intent Compiler, a natural language resource planner, picks the right tier and infrastructure, compiles a costed, resource-estimated spec, and matches it to capacity you already have. If nothing available fits, it can provision new infrastructure automatically. Each plan separates what you stated from what the compiler derived, explains the reasoning behind every derived value, and flags the assumptions that drive cost, so nothing is a black box.
For teams that want more control, the advanced options let them decide how much of the plan to define themselves:
State an intent
Describe the goal in one line. The compiler derives a costed spec, traces where each value came from, and waits for your confirmation before anything is submitted.
Declare a platform
Bring your own virtual cluster, runtime, and training configuration. Federator.ai Cortex compiles the pipeline and binds the resource footprint to the submission gate.
Reserve infrastructure
Engineers who prefer to set every parameter can use the Megatron-LM integration instead, choosing the model, parallelism, and hyperparameters directly while Federator.ai Cortex maps the job onto the right GPUs.
Wingman AI: A Copilot for Data Center Operators
Wingman AI is an AI copilot powered by a large language model and connected directly to Federator.ai Cortex. It acts as a semantic layer over the operational metadata the platform already collects, querying live data each time a question is asked, so operators can ask questions in plain language and get clear, rendered answers.
New users can ask how to complete a task in Federator.ai Cortex and get step-by-step guidance instead of searching through documentation. Experienced operators can ask about the fleet itself:
“Why is zone-02 running hot?”
“What is the cost impact of the last thermal emergency?”
“Compare current fleet utilization to last month.”
What Teams Gain
Every new site adds capacity, but it also adds complexity. AboveCloud keeps a growing fleet manageable by turning separate data centers into a single, coordinated resource, so each GPU, wherever it sits, contributes to the work that matters most.
Faster Time to First Job
New users can submit real training jobs on day one. The Intent Compiler turns a one-line description into a complete request, and Wingman AI answers questions about the platform as they come up.
One Path for Every Skill Level
Researchers describe goals in plain language, ML engineers tune parallelism by hand, and platform teams reserve raw capacity. Every path passes through the same submission gate in Federator.ai Cortex.
Cost Visibility Before Submission
Every compiled request includes its estimated resources and cost, and nothing runs until the user confirms it. Teams see what a job will consume before a single GPU is reserved.
Answers Grounded in Live Data
Wingman AI draws on the operational metadata Federator.ai Cortex already collects, so its answers reflect the current state of the fleet rather than generic advice or outdated documentation.
Frequently Asked Questions
What is the Intent Compiler in Federator.ai Cortex?
The Intent Compiler is a natural language resource planner in Federator.ai Cortex. It turns a plain-language description of an AI workload, such as model size, training data, GPU type, and deadline, into a costed, ready-to-submit resource plan. Federator.ai Cortex works out the GPUs, parallelism, memory, and storage the job needs, and the user reviews the plan before anything is submitted.
How does the Intent Compiler differ from the Megatron-LM integration in Federator.ai Cortex?
The Intent Compiler starts from a goal described in plain language and derives the configuration automatically. The Megatron-LM integration is for engineers who want to set the model, parallelism, and training hyperparameters themselves. Both produce a resource plan that Federator.ai Cortex places on the right GPUs.
What is Wingman AI in Federator.ai Cortex?
Wingman AI is an AI copilot powered by a large language model and connected directly to Federator.ai Cortex. It acts as a semantic layer over the operational metadata the platform already collects, querying live data each time a question is asked, so operators can ask questions in plain language and get clear, rendered answers.
How does Wingman AI help new Federator.ai Cortex users?
Wingman AI gives step-by-step answers to questions about using Federator.ai Cortex, so new users can learn the platform while they work instead of searching through documentation. This shortens the learning curve and helps an AI data center see value from Federator.ai Cortex sooner.
Does the Intent Compiler submit jobs automatically?
No. The Intent Compiler produces a costed plan that the user reviews and confirms before submission. Users can optionally allow Federator.ai Cortex to provision new infrastructure when no existing capacity fits the request.