About LucidOps
LucidOps builds custom applications, AI agents, and operational systems for our partners. We work closely with businesses to understand how they operate, connect their tools and data, and build software that makes their teams more effective.
We’re looking for a hands-on AI engineer who can build across the full system: the agent, the application around it, and the automated workflows connecting it to real business operations.
You should already use AI-assisted development as a core part of how you work. We want someone who can move quickly with these tools while taking ownership of architecture, code quality, testing, and production reliability.
What You’ll Do
Build production AI agents using LangGraph and LangChain, including tool use, state management, persistent execution, and human-in-the-loop workflows.
Develop applications around AI capabilities, working across backend services, APIs, databases, and user-facing interfaces.
Build automated AI workflows that connect business systems, process documents and communications, and take actions with appropriate validation and approval controls.
Work across models and providers, choosing the right model for each task based on measured quality, latency, cost, and privacy requirements.
Own evaluations and observability using LangSmith. Create evaluation datasets, inspect traces, compare approaches, and catch regressions before they reach production.
Make systems reliable, handling retries, partial failures, duplicate events, permissions, and recovery without losing work or taking unintended actions.
Work directly with the LucidOps team and partners to turn business requirements into shipped, maintainable software.
What We’re Looking For
Experience shipping AI-powered software beyond prototypes. You can explain what you built, how it behaves in production, and what you learned from its failures.
Hands-on experience with LangGraph, LangChain, and LangSmith. You understand orchestration, tool calling, context management, tracing, and debugging.
Practical eval experience. You’ve built representative test datasets, defined success criteria, and evaluated prompts, models, retrieval, or end-to-end agent behavior. You understand where deterministic checks, human review, and LLM-as-judge evaluations fit.
Strong software engineering fundamentals, with proficiency in Python and the ability to contribute to TypeScript/JavaScript applications.
Experience integrating APIs, databases, and external services, including authentication, webhooks, and asynchronous processing.
An established AI-assisted development workflow. You regularly use tools such as Claude Code, Codex, Cursor, or equivalents to implement, test, debug, and review code. You can independently verify and maintain what they produce.
A model-agnostic approach. You’re comfortable working across commercial and open-weight models rather than designing everything around one provider.
Good engineering judgment. You know when an agent is useful, when a deterministic workflow is better, and when a human needs to stay in the loop.
Clear communication and ownership. You can clarify ambiguous requirements, explain tradeoffs, and carry work from initial design through deployment and ongoing improvement.
Nice to Have
Experience with retrieval-augmented generation, document processing, structured extraction, and context engineering.
Familiarity with MCP, reusable agent tools, and multi-agent architectures.
Experience with React, PostgreSQL, Docker, CI/CD, and cloud deployments.
Experience building software for operational teams or integrating business platforms.
What Success Looks Like
You ship systems that people use and can trust. Agents complete useful work, applications make that work accessible, and evaluations show whether changes actually improve performance. You move quickly with AI assistance without leaving behind fragile code or unverified behavior.
Pay: $45.00 - $100.00 per hour
Work Location: Remote