Security Engineer - AI Red Teamer - Bangalore

CareerXperts Technologies Private Limited; Bengaluru; On-site; INR 1,200,000 - 2,000,000; Full time

Job summary

CareerXperts Technologies Private Limited in Bengaluru is seeking an AI Red Teamer to identify weaknesses in LLMs, AI agents, and RAG systems. You will craft attack paths beyond traditional application security and provide clear remediation guidance.

The role requires hands-on security testing, familiarity with OWASP LLM Top 10 and MITRE ATLAS, and the ability to communicate complex findings effectively. Join us to push AI security boundaries.

Qualifications

• Up to 5 years in AI security, red teaming, pentesting, AppSec, or security research.
• Strong grasp of LLM/GenAI architectures, agents, RAG, vector DBs, and tool calling.
• Hands-on knowledge of OWASP LLM Top 10 and familiarity with MITRE ATLAS.
• Strong web and API security skills.
• Python (or similar) for automating tests.
• Ability to build unconventional attack scenarios and communicate findings clearly.

Responsibilities

• Adversarial testing of LLMs, agents, RAG pipelines, APIs, and AI-enabled apps.
• Attacks mapped to the OWASP Top 10 for LLM Applications and MITRE ATLAS.
• Prompt injection (direct and indirect), jailbreaks, data leakage, excessive agency, insecure tool use.
• Attack chains across agents, plugins, MCP servers, and third-party integrations.
• Cross-tenant data exposure, RAG poisoning, and privilege-boundary violations.
• Repeatable PoCs and test cases, with clear findings and remediation guidance.

Skills

AI security

Red teaming

Pentesting

AppSec

Security research

Tools

OWASP LLM Top 10

MITRE ATLAS

Job description

We're looking for an AI Red Teamer to find, reproduce, and demonstrate security weaknesses in LLMs, AI agents, RAG systems, and AI-integrated platforms. You'll think creatively about attack paths that go well beyond traditional application security.

What You'll Own:

• Adversarial testing of LLMs, agents, RAG pipelines, APIs, and AI-enabled apps
• Attacks mapped to the OWASP Top 10 for LLM Applications and MITRE ATLAS
• Prompt injection (direct and indirect), jailbreaks, data leakage, excessive agency, insecure tool use
• Attack chains across agents, plugins, MCP servers, and third-party integrations
• Cross-tenant data exposure, RAG poisoning, and privilege-boundary violations
• Repeatable PoCs and test cases, with clear findings and remediation guidance

What you bring:

• Up to 5 years in AI security, red teaming, pentesting, AppSec, or security research
• Strong grasp of LLM/GenAI architectures, agents, RAG, vector DBs, and tool calling
• Hands-on knowledge of OWASP LLM Top 10 and familiarity with MITRE ATLAS
• Strong web and API security skills
• Python (or similar) for automating tests
• Ability to build unconventional attack scenarios and communicate findings clearly

Bonus: original AI/LLM attack research, MCP/agentic testing, bug bounty, CTFs, CVEs, conference talks or public research.

If you enjoy breaking AI systems in ways their builders never imagined, let's talk.