Principal DevSecOps Engineer - AI
Company: Finastra
Location: Bengaluru
Type: Full-time
Posted: 2026-08-20
About this role
# Who are we?
At Finastra, we’re a global leader in financial services software, dedicated to expanding access to financial services and shaping what’s next for the industry. Our technology powers mission‑critical solutions across Lending, Payments and Universal Banking, supporting over 7,000 customers, including 80% of the world’s top 50 banks, in more than 110 countries.
A senior technical SME responsible for enabling secure, automated delivery of machine learning and AI workloads. This role operates at the intersection of AI/ML engineering, product development, platform infrastructure, and cybersecurity—embedding security controls across the entire AI lifecycle, including CI/CD pipelines, model registries, vector databases, MCP services, and agent‑based AI orchestration.
Key Responsibilities:
Secure AI Platform Engineering
- Architect and implement secure, scalable, and resilient CI/CD pipeline’s purpose‑built for AI workloads, including LLM‑based and agentic AI systems.
- Automate secure build, deployment, and promotion workflows for AI workloads.
ML/AI SecOps Integration
- Shift security *left* by embedding automated controls such as SAST, SCA, container/image scanning, and policy‑as‑code into pipelines.
- Implement AI‑specific security measures, including data‑poisoning safeguards, model integrity validation, and prompt‑injection mitigation.
Infrastructure & Platform Security
- Work with platform engineering teams to harden vector databases, retrieval APIs, and other AI-serving infrastructure.
- Establish secure Infrastructure‑as‑Code (IaC) templates, GitOps workflows, and micro‑segmented identity patterns using secrets‑management tools such as Vault.
- Ensure multi‑region, fault‑tolerant, and security‑hardened deployment architectures for AI platforms.
Model Supply Chain Security
- Implement provenance, attestation, and traceability for datasets, training pipelines, and model artifacts.
- Protect the end‑to‑end AI supply cha...