About the role

Destinare Services is seeking Application Security Engineers to identify, reproduce, and assess vulnerabilities in applications powered by large language models (LLMs), agents, and connected tools.

You will review source code, trace data flows, assess access controls, and create reproducible tests for issues involving sensitive information disclosure, excessive agency, prompt injection, and tool abuse. Candidates need relevant security experience in at least one of Python, Go, or TypeScript/JavaScript.

Responsibilities

- Conduct secure code reviews of LLM applications, agent frameworks, MCP servers, and connected tools.
- Trace data flows across files, callbacks, registries, and background jobs to find routes that bypass security controls.
- Assess authentication, authorisation, least privilege, tenant isolation, trust boundaries, and approval workflows.
- Investigate sensitive information disclosure, excessive functionality, overly broad permissions, and unsafe agent autonomy.
- Write reproducible exploit tests and verify that patches prevent exploitation.
- Use CodeQL, Semgrep, SARIF results, and manual analysis to investigate vulnerabilities, including logic and authorisation flaws.
- Map confirmed findings to relevant CWE categories and document clear technical evidence.
- Review security patches and commit history to understand root causes and assess remediation.
- For reviewer-level assignments, lead code reviews and validate the accuracy and clarity of findings.

Requirements

- At least 3 years of experience in application security, secure code review, penetration testing, or vulnerability research.
- Reviewer-level candidates: at least 5 years of relevant experience, including leading code reviews.
- Solid source-to-sink data-flow reasoning and ability to explain how a security control can be bypassed.
- Understanding of authentication, authorisation, least privilege, tenant isolation, trust boundaries,



and approval workflows.
- Familiarity with the OWASP Top 10 for LLM Applications, especially Sensitive Information Disclosure and Excessive Agency, as well as prompt injection and tool abuse.
- Ability to map findings to CWE categories such as CWE-200/201, CWE-532, CWE-284/285/862/863, and CWE-250/269.
- Experience writing reproducible tests that distinguish vulnerable behaviour from patched behaviour.
- Experience with CodeQL or Semgrep, reading SARIF results, and understanding the limits of pattern-based scanners.
- Hands-on familiarity with LLM tool calling, agents, MCP, RAG, memory, and tracing. Relevant technologies include LangChain, LlamaIndex, OpenAI or Anthropic SDKs, and OpenTelemetry.
- Working knowledge of Docker and Git, and ability to review security patches and commit history.
- Clear written technical English.

Relevant experience in at least one language is required:

- Python: Security reviews involving unsafe deserialisation, subprocess and shell tools, logging and tracing leaks, asynchronous task queues, or dependency-injected authentication in FastAPI.
- Go: Security reviews involving MCP and agent tool servers, os/exec and file tools, context and middleware authentication, or broadly scoped Kubernetes or cloud credentials.
- TypeScript/JavaScript: Security reviews involving server actions and API routes, middleware authentication gaps, client-versus-server data exposure, MCP tool scopes, or OAuth token handling.

Nice to have

- OSWE, OSCP, GWAPT, CSSLP, or a similar security certification.
- CVE credits, bug bounty experience, or published security research.
- Experience with AI red-teaming or LLM evaluation.

Pay: Up to ₹1,500,000.00 per year

Application Question(s):

- How many years of experience do you have in application security, secure code review, penetration testing, or vulnerability research?
- Which of Python, Go, or TypeScript/JavaScript have you used for security reviews? Briefly describe relevant work

Work Location: Remote

📌 Application Security Engineer – LLM & AI Applications (India)
🏢 Destinare Services
📍 India