Job Description

Alice is building its internal AI infrastructure layer from the ground up. We have real agents running in production, a growing base of employees using AI in their daily work, and a clear architectural direction. What we don't have yet is a dedicated engineer to own it.

You'll be the first. Your job is to close the gap between "working prototype" and "production platform" - owning the foundation that hosts our agents, the pipelines that ship them, and the reliability layer (observability, cost controls, audit trails, evals) that makes it safe to run AI at scale in a trust & safety company.

This is an infrastructure-first role with deep AI fluency - not a prompt engineer, not a wrapper-framework operator, not a no-code builder. You should be equally comfortable writing a Terraform module, debugging a Kubernetes pod, and tracing an agent's tool-call chain.

We don’t operate with a predefined backlog here; you will be responsible for identifying high-impact needs and bringing them to life. The perfect fit for this role has a track record of deploying agentic systems that have held up under real-world usage, balances a focus on infrastructure with a deep concern for user experience, and recognizes that the primary hurdle in AI integration is rarely the model itself.

Responsibilities:

Platform & Infrastructure

• Architect, build, and run the AWS/Kubernetes platform that hosts Alice's internal AI agents and tools; drive AWS Well-Architected pillars (operational excellence, security, reliability, performance, cost, sustainability).
• Own Infrastructure-as-Code: Terraform modules, standards, and reviews for Bedrock, agent runtimes, vector DBs, and supporting services.

AI Systems

• Design and ship production-grade agents and multi-agent pipelines using the Anthropic Agent SDK, Claude Code, AWS Bedrock, and MCP — not wrapper frameworks.
• Own the full agent lifecycle: scoping → prototyping → eval → deploy → monitor → iterate.
• Integrate agentic workflows into internal and product systems via APIs, databases, webhooks, Slack, and email.

Reliability, Observability, Cost

• Build first-class observability across apps and infra: OpenTelemetry, Prometheus, plus LLM-specific tracing (Langfuse or equivalent), token/cost metrics, and eval pipelines.
• Define SLOs/SLIs and error budgets for AI services - latency, model fallback chains, eval regression gates, agent success rates. Lead incident readiness, response, and post-mortems.
• Drive FinOps: model routing by cost, cache hit rates, batch vs. realtime tradeoffs, budget alarms, per-team chargeback visibility.
• Implement guardrails: prompt-injection defenses, PII redaction, model allowlists, human-in-the-loop checkpoints, audit trails.

Org Impact

• Identify high-leverage workflows across the organization and translate them into scalable agentic automations.
• Partner with R&D, Delivery, security, and external vendors to deliver platform capabilities.

Requirements

Requirements (must-have)

• 3-5 years in software engineering, shipping and operating production-grade systems.
• 2+ years hands-on AWS, Kubernetes, and Terraform in production — not familiarity, ownership.
• 1-2 years hands-on building and deploying LLM-powered or agentic systems in production.
• Proficiency in Python: async patterns, REST APIs, cloud-native architecture.
• Production experience with native agentic SDKs (Anthropic Agent SDK, Claude Code) and MCP - tool-calling patterns, server configuration, memory systems, vector DBs.
• Hands-on AWS Bedrock for model access, IAM-based auth, and enterprise deployment patterns.
• Production CI/CD ownership (GitHub Actions, Argo CD, or equivalent) and observability stack experience (OpenTelemetry + Prometheus, plus LLM tracing).
• Proven ownership: design → implement → release → operate → improve, independently and within a team.
• Strong debugging instincts across multi-step agent chains and distributed infrastructure.
• Clear written and verbal communication in English; comfortable with internal and external stakeholders.
• Startup mindset: move fast, own decisions end-to-end, comfortable with ambiguity.

Nice to Have

• Background in trust & safety, content moderation, or compliance-sensitive environments.
• FinOps experience at scale (cost attribution, budget enforcement, optimization playbooks).
• Experience building lightweight internal dashboards or UI layers for agentic workflows.
• LLM evaluation framework experience (Braintrust, Langfuse evals, custom harnesses).

About Alice

Alice is a trust, safety, and security company built for the AI era. We safeguard the communicative technologies people use to create, collaborate, and interact—whether with each other or with machines.

In a world where AI has fundamentally changed the nature of risk, Alice provides end-to-end coverage across the entire AI lifecycle. We support frontier model labs, enterprises, and UGC platforms with a comprehensive suite of solutions: from model hardening evaluations and pre-deployment red-teaming to runtime guardrails and ongoing drift detection.