Agent Engineering Intern

Job Title: Agent Engineering Intern

Location: Bangalore / Hybrid

Company: DecisionX

Type: Paid Internship | Full-Time | 36 Months

About DecisionX

DecisionX is a Decision AI platform for Strategy and Analytics teams.

Think of it as the decision infrastructure layer for the modern enterprise.

Not a dashboard. Not a copilot. A system that governs how organisations reason, act, and learn from outcomes. Our customers include leading consumer businesses youve definitely heard of.

Were a small team working on problems that havent existed before. Our platform runs on a cognitive ontology engine - a structured graph of decisions, actions, goals, and outcomes that governs how an organisation reasons. Making this engine reliable across domains, industries, and edge cases is an unsolved problem. This role works directly on that layer.

Role Overview

This is not an application engineering role.

You will be working on reasoning infrastructure the layer beneath the product, not the render layer.

The core challenge:

How do you make a system reason consistently, correctly, and contextually across completely different business environments?

This role sits at the intersection of:

• ontology design
• evaluation systems
• LLM reasoning behaviour
• and real-world business logic

What Youll Work On

• Design and run evaluation frameworks across domains FMCG, Pharma, D2C, BFSI
• Build systems to detect when reasoning breaks, and why
• Audit and improve ontology coverage
• Identify gaps in cognitive node taxonomy and test whether system representations map to real business workflows
• Document experience gaps across prompts and UI
• Trace where outputs drift from intent and where clarity breaks
• Build lightweight internal tooling
• For eval tracking, ontology diffing, and coverage reporting
• (Tools the team actually uses not dashboards that collect dust)
• Work on the reconciliation layer
• The interface between unstructured policy documents and structured transaction data
• Test how the system handles conflict, ambiguity, and edge cases

What Were Looking For

• You think clearly about systems
• Not just code but the semantics of what the code represents
• Comfortable working in LLM-adjacent infrastructure
• (eval design, prompt engineering, structured outputs)
• You work empirically
• Form hypotheses, test them, and report results cleanly
• You see error modes as signals of deeper architectural gaps, not isolated bugs
• Familiarity with knowledge graphs, ontology design, or reasoning systems is a plus
• Backgrounds in CS, cognitive science, or linguistics all fit
• as long as you can think across layers
• You are curious about how organisations think, not just how software runs

What Youll Get

• Direct exposure to reasoning system design at the frontier of AI
• The opportunity to work on core architecture, not just surface features
• A front-row seat to building a new category in decision intelligence
• Ownership of problems that dont have existing play books
• Work that compounds into deep technical and conceptual leverage