AI Evaluation Engineer / Trust Engineer

The Ksquare Group; Hyderabad; On-site; INR 900,000 - 1,500,000; Full time

Job summary

The Ksquare Group is seeking a skilled QA engineer to join the AI evaluation and security testing team in Hyderabad. You will build evaluation datasets, run automated and manual testing, and integrate quality gates into CI/CD.

Responsibilities include testing agent access controls, configuring guardrail policies, and performing AI supply-chain checks while documenting results for compliance. The role requires strong Python/TypeScript testing, knowledge of eval frameworks, and hands-on experience

Qualifications

• Strong hands-on Python and TypeScript test engineering, including PyTest, Playwright, UI/API/integration testing.
• Eval frameworks (DeepEval, Ragas, Promptfoo, LangSmith or similar); LLM-as-judge.
• Prompt and model testing; CI/CD.
• AI security testing fundamentals: OWASP Top 10 for LLM Applications, prompt-injection and jailbreak techniques.
• Working knowledge of authentication and authorization for APIs and agents (OAuth 2.0, API keys, secrets management).
• Hands-on AI and agent workflow understanding, with demonstrated skill building and lifecycle management for application testing and AI evaluation.
• Observability and data analysis.

Responsibilities

• Build evaluation datasets, golden sets and rubrics for agents and RAG systems.
• Implement automated evals: groundedness, accuracy, tool-call correctness, safety, bias, latency and cost.
• Run automated and manual red-teaming of agents and RAG systems: prompt injection (via documents, web content, emails and tool outputs), jailbreaks, tool misuse, excessive agency, privilege escalation and sensitive-data leakage, using tools such as PyRIT, garak or Promptfoo.
• Test agent access controls: verify least-privilege tool and data permissions, MCP and connector authorization scopes, human-approval checkpoints and completeness of audit logs.
• Configure and validate AI gateway and guardrail policies (sensitive-data redaction, content safety, topic restrictions, rate and cost limits); measure false-positive and false-negative rates and tune policies with platform teams.
• Run AI supply chain checks: scan models, datasets and AI dependencies, and maintain AI bill of materials (AI-BOM) records.
• Wire eval suites into CI/CD as quality gates; track regressions across prompt and model versions.
• Monitor production quality, drift, guardrail effectiveness, cost and latency; support AI incident triage and report trust metrics to stakeholders.
• Build reusable agentic testing and evaluation skills for test design, test data preparation, UI/API/integration testing, execution, defect triage and AI evaluation; package prompts, scripts, harnesses and rubrics for project-team use.
• Manage testing skill catalogs, versions, dependencies, benchmark validations, releases and retirement; document usage, onboard project teams and incorporate feedback, with human review and traceable evidence for AI-generated tests and results.

Skills

Python

TypeScript

PyTest

Playwright

UI testing

API testing

Test harness

Eval frameworks

LLM as judge

CI/CD

AI security

OAuth 2.0

Secrets management

Observability

Data analysis

Tools

PyRIT

garak

Promptfoo

LangSmith

Job description

The Ksquare Group is a global technology consulting company headquartered in Dallas, Texas, with over 14 years of experience delivering technology solutions across Salesforce, cloud, data, AI, and network engineering for clients in Banking & Financial Services, Public Sector, Telecommunications, Manufacturing, Logistics, Healthcare, and Enterprise Services. With a team of 350+ professionals across the US, Mexico, the Dominican Republic and India, we help organizations solve complex business challenges and achieve measurable outcomes through our collaborative POD-based delivery model. At The Ksquare Group, we foster a culture of integrity, collaboration, continuous learning, and professional growth, empowering our teams to make a meaningful impact for clients worldwide.

Role purpose

Enable project teams to test and secure applications and AI agents by building and managing reusable QE, AI evaluation and AI security testing skills, test harnesses and quality gates. Combine application-stack and test engineering proficiency with AI and security expertise to assess functional quality, accuracy, safety, security, reliability and cost before and after release, and run continuous assurance for managed-service clients.

Key responsibilities

• Build evaluation datasets, golden sets and rubrics for agents and RAG systems.
• Implement automated evals: groundedness, accuracy, tool-call correctness, safety, bias, latency and cost.
• Run automated and manual red-teaming of agents and RAG systems: direct and indirect prompt injection (via documents, web content, emails and tool outputs), jailbreaks, tool misuse, excessive agency, privilege escalation and sensitive-data leakage, using tools such as PyRIT, garak or Promptfoo.
• Test agent access controls: verify least-privilege tool and data permissions, MCP and connector authorization scopes, human-approval checkpoints and completeness of audit logs.
• Configure and validate AI gateway and guardrail policies (sensitive-data redaction, content safety, topic restrictions, rate and cost limits); measure false-positive and false-negative rates and tune policies with platform teams.
• Run AI supply chain checks: scan models, datasets and AI dependencies, and maintain AI bill of materials (AI-BOM) records.
• Wire eval suites into CI/CD as quality gates; track regressions across prompt and model versions.
• Monitor production quality, drift, guardrail effectiveness, cost and latency; support AI incident triage and report trust metrics to stakeholders.
• Build reusable agentic testing and evaluation skills for test design, test data preparation, UI/API/integration testing, execution, defect triage and AI evaluation; package prompts, scripts, harnesses and rubrics for project-team use.
• Manage testing skill catalogs, versions, dependencies, benchmark validations, releases and retirement; document usage, onboard project teams and incorporate feedback, with human review and traceable evidence for AI-generated tests and results.
• Produce responsible-AI documentation and audit-ready compliance evidence, mapping evaluation and security test results to controls in NIST AI RMF, ISO/IEC 42001 and client policies.

Required skills

• Strong hands-on Python and TypeScript test engineering, including PyTest, Playwright, UI/API/integration testing and reusable test harness development; ability to understand and debug application flows, APIs and data dependencies.
• Eval frameworks (DeepEval, Ragas, Promptfoo, LangSmith or similar); LLM-as-judge.
• Prompt and model testing; CI/CD.
• AI security testing fundamentals: OWASP Top 10 for LLM Applications, prompt-injection and jailbreak techniques, and hands-on use of at least one red-teaming tool (PyRIT, garak, Promptfoo or similar).
• Working knowledge of authentication and authorization for APIs and agents (OAuth 2.0, API keys, secrets management).
• Hands-on AI and agent workflow understanding, with demonstrated skill building and lifecycle management for application testing and AI evaluation, including prompt/tool design, packaging, version control, validation and project-team enablement.
• Observability and data analysis.

Preferred

Application or cloud security testing background (e.g., OSCP, CEH or a cloud security certification); MITRE ATLAS; experience with AI gateways or guardrail frameworks (e.g., NeMo Guardrails, Llama Guard); knowledge of NIST AI RMF and ISO/IEC 42001.

Success measures

Testing and evaluation skills adopted across projects, reuse and test cycle efficiency, application and eval coverage, defects caught pre-release, regression escape rate, trust metrics reported; security vulnerabilities caught before release; guardrail precision and recall; managed-assurance SLAs met.