CareerPlanSign in

Software Engineer - AI

USA💼 Full-time💰 $90,000–$90,000🗓 2026-05-21 → 2026-09-26

Core

Human-in-the-loop training and evaluation of advanced generative AI agents and large language models to improve their autonomy and multi-step workflow capabilities.

Role type

AI Training & Evaluation Engineer

Builds

Proactive, multi-step autonomous agents and complex real-world architectural workflows for leading AI organizations

Domain

Generative AI, Autonomous Agents, Workflow Orchestration

Deliverable

production ML models

Required skills

Backend engineering, AI automation, complex systems integration, Python, JavaScript, Go, Java, SQL, multi-stage coordination workflows, agent-to-tool/API integration, persistent state management, session tracking, security vulnerability identification (privacy exposure, authority escalation, prompt injection)

Preferred skills

Experience connecting agents to real tools and APIs (e.g., Supabase, Gmail), familiarity with MEMORY.md for agent progress monitoring

Technologies

Java, JavaScript, Python, SQL

Responsibilities

Train advanced generative systems by providing human feedback to improve AI agents, collaborate with AI teams to teach LLMs to operate as proactive multi-step agents, design and refine complex real-world architectural workflows, evaluate and improve system behavior across multi-step interactions in live environments, provide detailed technical observations to support autonomous system development, identify and diagnose nuanced failures in agent behavior including security and privacy concerns

Sourced via devitjobs · Listed on CareerPlan, which tracks 70,000+ jobs from 20+ sources.