AI Agent Engineer

Location: Remote

Type: Full-Time

Reports to: Head of R&D

Rexi Group is building a next generation AI platform and we are looking for a strong software engineer who can design and build production grade AI agents entirely in code. This is not a no code automation role. We are not looking for someone who wires together drag and drop workflows.

We want someone who understands programming deeply. Architecture. Data flow. System behavior. Reliability. Someone who can build intelligent agents that operate consistently in real world environments.

What You’ll Do

Architect and implement multi-step AI agents in production code
Design reasoning loops, tool orchestration layers, memory systems, and fallback strategies
Engineer robust structured data pipelines
Build scraping and browser automation systems
Integrate directly with LLM APIs (OpenAI, Anthropic, etc.), managing:

Prompt architecture

Tool schemas

Streaming responses

Error handling

Cost optimization

Design systems that handle probabilistic outputs with deterministic safeguards
Implement monitoring, logging, evaluation, and feedback loops
Collaborate on long-term system architecture and scalability

What We’re Looking For

Core Requirements

Strong programming foundation in a serious language (Python, Go, Rust, Java, C#, TypeScript, etc.)
Deep understanding of:

Data structures

Clean architecture

Modular system design

Asynchronous workflows

Experience building real production systems (not just prototypes or scripts)
Solid understanding of databases (relational and non-relational)
Experience working with APIs and distributed systems
Ability to design reliable logic around non-deterministic outputs
Comfort debugging complex, multi-component systems

Nice to Have

Experience with LangGraph, LangChain, or custom agent frameworks
Experience with vector databases
Experience with Playwright, Puppeteer, Selenium, or large-scale scraping systems
Experience deploying and maintaining production systems (Docker, cloud infra, CI/CD)
Experience building multi-agent or tool-using systems

What We Value