Job Description

Role Purpose

The purpose of this role is to analyse, identify, rectify & recommend specific improvement measures that help in the security posture of the organization by protecting the sensitive information
 

͏

Do

• ML/AI Engineer – AI Red Team & Security Testing (x4 Positions)

  Experience: 5–8 years

  Location: Bangalore only 

  
  Role Overview:
  
  

  As an ML/AI Security Engineer, you will build, configure, and operate an open-source adversarial testing framework within a fully air-gapped banking environment. You will execute red team exercises against three on‑prem LLMs, generate synthetic attack payloads, score results, and produce comprehensive technical reports. This is a hands-on, engineering-heavy role working with PyRIT, Ollama, CyberSecEval, and related tooling.

   

  🎯 Key Responsibilities

   

  • Deploy & configure open-source AI security testing stack (PyRIT, Ollama, CyberSecEval, HarmBench) on air-gapped/private cloud infrastructure (NVIDIA B300 GPUs).
  • Execute adversarial red team campaigns against 3 on-prem LLMs using 100% synthetic test data.
  • Map & validate 5 attack types to MITRE ATLAS & OWASP LLM Top 10.
  • Operate attacker & judge/scoring LLMs to automate prompt generation, exploit execution, and result scoring.
  • Produce industry-standard penetration testing reports: Executive Summary, Scope & Method, Technical Narrative, Findings, Recommendations.
  • Author installation guides, architecture diagrams, SOPs, and sample red team reports for bank handover.
  • Conduct knowledge transfer sessions enabling U.S. Bank teams to independently run future tests.
  • Deploy & configure open-source AI security testing stack (PyRIT, Ollama, CyberSecEval, HarmBench, vLLM) on bank-provisioned NVIDIA B300 GPU infrastructure in an air-gapped/VDI environment.
  • Execute adversarial red team campaigns against 3 on-prem LLMs, testing raw baselines without guardrails using automated prompt generation and human-guided multi-turn probing

͏

Deliver

͏

͏