PRX – AI Engineer

Role Overview

Goldman Sachs is seeking an experienced PRX – AI Engineer to design, build, and deploy GenAI and Agentic AI solutions that improve reliability, reduce risk, and lower operational costs in large-scale production environments. This role focuses on building intelligent agents capable of diagnosing issues, reasoning over complex systems, and executing secure actions in production.

Note: No low-code / no-code experience is applicable for this role. Strong hands-on engineering is mandatory.

Key Responsibilities

Agentic AI & GenAI Development

• Design and implement agentic AI systems using tool-calling agents, retrieval, structured reasoning, and secure action execution.
• Implement function calling, policy enforcement, and orchestration following MCP protocols.
• Engineer strong guardrails for safety, compliance, and least-privilege access.

LLM Production & Evaluation

• Productionize LLMs by building evaluation frameworks for open-source and foundational models.
• Develop retrieval pipelines, prompt orchestration, response validation, and self-correction loops.
• Optimize LLM workflows for production operations and runtime environments.

Runtime & Ecosystem Integration

• Integrate AI agents with observability, incident management, CI/CD, and deployment systems.
• Enable automated diagnostics, runbook execution, remediation, and post-incident summaries with full auditability.

Collaboration & Business Impact

• Partner with production engineers and application teams to translate operational pain points into agentic AI solutions.
• Define success metrics aligned with reliability, risk reduction, and cost optimization.
• Deliver traceable and auditable outcomes.

Safety, Governance & Reliability

• Implement validator models, adversarial testing, policy checks, and deterministic fallbacks.
• Build circuit breakers, rollback mechanisms, and continuous evaluation pipelines.

Performance & Scalability

• Optimize latency and cost through prompt engineering, caching, model routing, batching, streaming, and parallel tool calls.
• Ensure systems meet strict production SLOs.

RAG & Knowledge Systems

• Design and maintain RAG pipelines with domain knowledge curation.
• Build data quality validation frameworks and feedback loops to ensure knowledge freshness.

Engineering Excellence

• Lead design reviews and enforce high-quality engineering standards.
• Mentor peers on agent architectures, evaluation methods, and safe AI deployment practices.

Essential Skills & Experience

• 5+ years of software development experience (Python preferred; Java, Go, C/C++ acceptable).
• 3+ years designing and launching production ML systems, including deployment, monitoring, evaluation, and data pipelines.
• Hands-on experience with LLMs, including:
  • API integrations
  • Prompt engineering
  • Fine-tuning / adaptation
  • RAG pipelines
  • Tool-using / function-calling agents
• Strong understanding of commercial and open-source LLMs (OpenAI, Gemini, Llama, Qwen, Claude, etc.).
• Solid foundation in applied statistics, ML concepts, algorithms, and data structures.
• Excellent problem-solving skills with strong ownership and communication abilities.

Preferred Skills

• Experience with AWS cloud infrastructure, including:
  • ECS / EKS, Lambda
  • S3, DynamoDB, Redshift
  • Step Functions
  • SageMaker
• Infrastructure as Code using Terraform or CloudFormation.
• Experience operating AI systems in large-scale production environments.