Full-time · Remote (India)

AI Reasoning Systems Architect

Cehpoint is hiring for this role: full-time, Remote (India), ₹15,00,000 – ₹25,00,000 / year + Performance Bonus. Pioneer tree-of-thought exploration, formal logic verifiers, automated theorem proving, and self-correcting reasoning loops for complex analytical domains.

About the role

Next-frontier AI requires verified reasoning over statistical pattern matching. In this role, you will build algorithmic search and verification pipelines (Tree-of-Thoughts, Graph-of-Thoughts, and verifier-guided decoding) that empower Cehpoint's agents to solve complex math, code, and architectural challenges with mathematical guarantees.

Lead the development of next-generation reasoning architectures. You will construct verifiable multi-step reasoning engines that dismantle ambiguous technical problems into formal proofs, eliminating hallucinated steps through automated verification.

What you will do

  • Design guided tree search algorithms (MCTS, beam search with learned value functions) for multi-step reasoning tasks.
  • Implement automated step-by-step logic verifiers that detect deduction flaws in real time.
  • Build recursive self-refinement loops that allow agents to debug their own intermediate solutions prior to output generation.
  • Optimize token-level reasoning traces to maximize cognitive density while minimizing inference expenditure.
  • Integrate formal verification engines (Z3 SMT solver, Lean, AST checkers) directly into the agent decision loop.

What you should know

  • Formal logic, discrete mathematics, proof theory, and computational complexity.
  • Thorough grasp of modern reasoning paradigms: Chain-of-Thought, Self-Consistency, Tree of Thoughts, and Process Reward Models (PRMs).
  • Solid computer science foundation in graph algorithms, search heuristics, and state space navigation.

What you should be able to do

  • Advanced Python and/or C++/Rust for high-speed search tree execution.
  • Demonstrated experience fine-tuning and prompting frontier models for structured reasoning tasks.
  • Experience integrating formal solvers (Z3, CVC5) or deterministic grammar validators.

How you work

  • Passion for mathematical rigor, formal proofs, and eliminating cognitive hallucinations.
  • Persistent drive to discover how synthetic models organize internal reasoning steps.
  • Pragmatic builder who values verifiable correct answers over eloquent fabrications.

Skills we check

  • Tree-of-Thought Search
  • Formal Logic Verification
  • Chain-of-Thought Optimization
  • Self-Refine Loops
  • Python / C++
  • Automated Theorem Proving
  • Reasoning Traces

What you get

  • Premium compensation (₹15,00,000–₹25,00,000) with quarterly performance bonuses.
  • Direct leadership of core intellectual property and flagship reasoning benchmarks.
  • 100% remote flexibility with home office infrastructure setup.
  • Full access to cutting-edge research clusters and frontier model reasoning APIs.

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: Develop a self-verifying reasoning pipeline that breaks down ambiguous mathematical/legal problems into verifiable lemmas, validating each step before producing a synthesized solution (strictly 5 working days).

  5. Offer

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