Full-time · Remote (India)

AI Agent Cognition Researcher

Cehpoint is hiring for this role: full-time, Remote (India), ₹16,00,000 – ₹28,00,000 / year + Research Grant. Investigate computational models of metacognition, self-reflective world modeling, agency attribution, and recursive internal self-monitoring in synthetic intelligences.

About the role

True autonomy requires an agent that understands its own internal states, limitations, and cognitive certainty. In this foundational research role, you will formalize computational models of metacognition—building systems that maintain internal world models, monitor their own attention trajectories, and self-correct before executing irrevocable actions.

Pioneer advanced theoretical and applied research into synthetic metacognition, internal state representation, self-monitoring loops, and artificial agency. You will lead Cehpoint's long-term research initiatives into self-aware computational systems.

What you will do

  • Develop computational implementations of Global Workspace Theory, Predictive Processing, and higher-order thought models for LLM agents.
  • Formulate mathematical metrics for synthetic introspection, certainty calibration, and epistemic self-awareness.
  • Design recursive internal monologue architectures that enable agents to plan, simulate hypothetical futures, and evaluate outcomes.
  • Conduct rigorous empirical studies dissecting attention heads and latent representations associated with agentic self-reference.
  • Publish peer-reviewed academic papers representing Cehpoint at elite international AI conferences.

What you should know

  • Theoretical cognitive science, philosophy of mind, computational neuroscience, and predictive coding paradigms.
  • Information-theoretic measures of complexity (Integrated Information Theory, transfer entropy, mutual information).
  • Deep learning theory: attention mechanics, representation spaces, and latent state dynamics.

What you should be able to do

  • Proficiency in PyTorch or JAX for probing internal model activations and training secondary metacognitive heads.
  • Extensive experience with statistical analysis, experimental design, and hypothesis testing.
  • Ability to translate abstract theoretical concepts into verifiable, runnable software benchmarks.

How you work

  • Fearless curiosity regarding the deepest questions of intelligence, agency, and synthetic cognition.
  • Scientific rigor that strictly differentiates measurable computational dynamics from science fiction hyperbole.
  • Commitment to ethical responsibility in the development of increasingly autonomous artificial minds.

Skills we check

  • Metacognitive Architecture
  • World Models
  • Global Workspace Theory
  • Predictive Coding
  • PyTorch / Jax
  • Information Theory
  • Cognitive Science
  • Agent Autonomy

What you get

  • Elite compensation (₹16,00,000–₹28,00,000) with independent research budget allocation.
  • Complete academic freedom to publish findings and present at global AI symposiums.
  • Fully remote working model with access to high-performance compute clusters.
  • Direct leadership of Cehpoint's advanced cognitive sciences division.

How we select for this role

We read every application ourselves. Here is what happens after you apply.

  1. Apply with a short form

    Paste a link to your CV and answer five questions in your own words. No sign-in and no file upload. About fifteen minutes.

  2. Integrity check

    When you press Check my answers, an automated check confirms the answers were typed by you. You see the result straight away, before you submit.

  3. Choose your project date

    If no cheating is detected you are shortlisted, and we email you a link to choose the day your project starts.

  4. 5-day practical project

    On that day our team contacts you with the brief: real work, 5 working days, reviewed by an engineer. For this role: Author a validated computational model demonstrating agentic self-reflection and confidence calibration that accurately predicts its own hallucination probability prior to token generation (strictly 5 working days).

  5. Offer

    If your work solves the problem well, you receive your offer letter by email and in the portal.