AI Engineer Intern

About the Role

As an AI Engineer Intern, you will work full-time on-site, embedded directly with the founder on core product and security features. This internship involves shipping into a live, adversarially-tested system.

Key Responsibilities

• Building and refining rules in the OPA/Rego policy engine that governs agent actions.
• Working on the LLM-based semantic risk scoring system that evaluates agent actions for intent and harm potential.
• Contributing to taint tracking — tracing how untrusted or sensitive data flows through an agent's tool calls.
• Helping design and test human-in-the-loop approval workflows.
• Working with the MCP protocol layer that enables transparent agent-tool communication.
• Running adversarial tests against the system to find and close security gaps.
• Contributing to Indic-language-aware risk detection, using Sarvam AI's models/credits.
• Writing documentation, test cases, and audit trail features.

Requirements & Tech Stack

• Solid Python fundamentals.
• Understanding of LLMs and how agentic systems work (tool calling, multi-step agent loops, RAG, frameworks like LangChain/LangGraph/CrewAI), or willingness to learn fast.
• Genuine interest in security/adversarial thinking — comfortable trying to break a system.
• Familiarity with or eagerness to learn OPA/Rego (policy-as-code).
• Understanding of MCP (Model Context Protocol), or ability to ramp up quickly.
• Backend/API experience.
• Comfortable in a fast-moving, early-stage environment.

Qualifications

• Any CTF, bug bounty, or vulnerability-research experience is a strong plus.
• Bonus: open-source contributions, especially anything security-related or published packages.