Staff Software Engineer - AI Applied Engineering
Indexed description
You would work across the whole loop with the team. We shape the idea, build the agents, ship them to users, see how they hold up with real work, and come back to make them better. We also write and own the evals that tell us whether the product is any good, so quality stays with the engineers building it.
We are looking for builders who already build with AI every day. You use coding agents in your own workflow, and you have shipped LLM-backed features to production and lived with them afterward, so you know how to keep them reliable for real users. You would join a team that stays close to its users and holds a high bar on quality. We take ownership together.
What You Will Do
- Build and ship fullstack features end-to-end, from design through production
- Extend our Claude Agent SDK integration MCP tool design, streaming (SSE) chat, context compaction, tool-permission gating
- Design and implement LLM-based features (RAG, agents, tool use) that power agents generating code, configuring workflows, and importing data
- Build reliable AI pipelines with a focus on latency, cost, and failure modes
- Define and implement evaluation strategies for AI outputs: offline (curated test sets, regression checks before shipping) and online (live production metrics: accuracy, override/escalation rates, latency, cost)
- Own prompt/context engineering strategy and drive model and approach iteration
- Design guardrails, tool-permission gating, and fallback mechanisms for non-deterministic outputs
- Strong React/TypeScript experience, with a track record of shipping production features
- Comfortable working across frontend and backend
- Experience building complex, stateful UI - visual/node-based editors, real-time or streaming interfaces, or similar
- Strong architectural instincts, considers system and product impact when making decisions
- Hands-on experience with LLM applications: RAG, agents, tool use, or similar, ideally with the Claude API/Agent SDK or MCP tool servers specifically
- Practical experience using LLMs in real production systems, including streaming responses and prompt/context engineering
- Ability to reason about system design, not just about models or prompts
- Demonstrated ability to move a feature from prototype to production, including monitoring, cost control, and failure handling
- A pragmatic engineering mindset - knows when not to use AI
- Comfort working in a fast-moving, early-stage product environment
- Comfortable using AI coding assistants (Copilot, Cursor, Claude Code, or similar) to speed up day-to-day development
- Some exposure to AWS-based backends (DynamoDB, Lambda, ECS)
- Experience with search, recommendation systems, or other data-heavy systems
- Experience with observability/monitoring of production systems
- Familiarity with evaluating AI/ML features in production (offline eval sets, online metrics, A/B testing)
- Exposure to durable/stateful execution engines (Temporal, AWS Step Functions/Durable Execution, or similar)
- Experience with vector databases
- Production ML or MLOps experience
- Bun
- Claude Agent SDK
- MCP (Model Context Protocol)
- Vercel AI SDK
- RAG
- Vector DBs
- AWS
Benefits
- Unlimited remote work policy
- 25 days of annual leave, 2 additional days off for volunteering
- Competitive remuneration package incl. an annual bonus
- Top-notch IT setup.
- Mental health support, life insurance, food vouchers
- Additional health insurance - for you and your loved ones
- Annual wellness support allowance
- External Professional Learning Opportunities
Create a free Caio profile to unlock more results and save your role and location preferences.
Unlock free search