Software Engineer III, Applied Agentic AI

This is a fully remote position, open to applicants in Argentina.

📋 Description

• Develop reference agentic applications on Proofpoint's platform, focusing on the design, evaluation, deployment, and monitoring of agents.

• Create and document best practices for tool design and MCP integration, including prompt and context management, memory and planning, cost/latency tradeoffs, failure recovery, and release readiness.

• Design evaluation harnesses featuring task suites, baselines, and metrics for quality, cost, and latency, utilizing them to determine what gets released.

• Transform ambiguous inquiries into functional prototypes swiftly and present findings to the requesting teams.

• Advise on when to discontinue an approach.

• Refine successful prototypes into reusable platform capabilities for product teams.

• Collaborate with seasoned engineers, security researchers, and product tech leads.

• Disseminate agentic patterns across the organization through demonstrations, design reviews, written documentation, and pairing sessions.

• Communicate and comprehend stakeholder requirements while engaging in cross-team design discussions.

⛳️ Requirements

• Bachelor’s or Master’s Degree in Computer Science or a related field, or equivalent practical experience.

• Over 3 years of engineering experience, with recent hands-on involvement in LLM or agentic systems that have progressed beyond the prototype phase.

• Strong understanding of the agentic development lifecycle and its elements, including agent runtimes, tool and MCP integration, prompt and context management, memory, planning, orchestration, evaluations, tracing, guardrails, as well as rollout and rollback processes.

• Proficient in Python; knowledge of TypeScript/Node is a plus.

• Experience collaborating with globally distributed teams.

• Background in building and managing services within a cloud microservices environment, including containers, CI/CD, and automated deployment; AWS experience is a plus.

• Comfortable diagnosing distributed systems using traces and metrics; familiarity with OpenTelemetry (including GenAI semantic conventions) is advantageous.

• Possess an evaluation-first mindset: measurement is integrated into the build process, not conducted afterward.

• Genuine enthusiasm for cutting-edge AI, with experience experimenting with agent frameworks and LLM SDKs like LangGraph, Anthropic/OpenAI Agents SDK, MCP, LangChain, or similar technologies.

• Proactive and inclined toward execution — capable of tackling ambiguous problems, developing a Proof of Concept in days instead of quarters, and continuing progress despite unclear pathways.

• Quick learner with a genuine curiosity about how underlying systems function, beyond just their usage.

• Preferred: Experience with AI coding tools (such as Claude Code, Codex, or similar) and familiarity with agent interoperability protocols like MCP and A2A.

• Preferred: Experience in building evaluation harnesses, benchmarks, or ground-truth datasets.

• Preferred: Creating developer-facing reference implementations, SDK samples, or internal training materials.

• Preferred: Knowledge of AG and semantic retrieval (vector stores, hybrid search), along with experience working with both hosted APIs and open-weight models.

🏝️ Benefits

• Competitive compensation

• Comprehensive benefits

• Career success on your own terms

• Flexible work environment

• Annual wellness and community outreach days

• Continuous recognition for your contributions

• Global collaboration and networking opportunities