AI Agent Engineer (Agentic Automation)
Scraped from: Afriworket
AI Agent Engineer (Agentic Automation) zeami.io [ Zeami.io ] · [Addis Ababa / Hybrid] · [Full-time] About us We build an AI platform that observes how teams work, identifies their repetitive workflows, and automates them with AI agents. Our platform already surfaces thousands of automation candidates and flags the ones ready to build the bottleneck is turning those into deployed, reliably-running agents. That's the role. The role You own the "automate" half of the pipeline: take a validated workflow, build it into a production AI agent, wire it into the target systems (ERP, WhatsApp Business, accounting, in-house apps), deploy it, and harden it until it reliably does the work. You'll also strengthen the build engine itself so agents ship faster and break less. This is a deep, hands-on engineering role you're measured by agents that run correctly in production, not by meetings. What you'll do • Build AI agents from ready workflow candidates multi-step logic, tool/function calling, guardrails, retries. • Build and maintain connectors to the systems agents need to drive (APIs, webhooks — and RPA/computer-use when there's no API). • Deploy, monitor, and harden agents in production until they reliably replace the manual work. • Improve the agent build/runtime itself reliability, cost, latency, evaluation. • Debug across the full pipeline: trace a misbehaving agent back through the workflow data to root cause. • Feed quality issues back to the core team (duplicated/bad workflows, wrong cost/ROI signals). Must-have • Strong Python in a real backend codebase (FastAPI + PostgreSQL). • Hands-on LLM / AI agent engineering prompting, tool/function calling, and making non-deterministic systems behave reliably in production. • Agent frameworks LangGraph / LangChain (our agent runtime). • Building integrations/connectors REST, auth, webhooks; reverse-engineering a system without clean docs. • RPA / computer-use / browser automation for systems with no API. • Agent orchestration (Prefect or equivalent durable, resumable jobs). • Solid SQL and root-cause debugging discipline. • Awareness of multi-tenant correctness (data isolation across clients). • Vision / multimodal LLM experience. • LLM cost/latency optimization and agent evaluation/eval-harness work. • AWS / Docker • Shipped production AI agents at scale, or strong agent side-projects.