Supply Chain Operations Task Designer (Python)

Quick Facts

• Task-based compensation model (paid per task)
• Tasks take ~60–75 minutes each; vetting is done via tasks (not live interview)
• Work focuses on creating benchmark tasks to evaluate AI agents using realistic supply chain planning scenarios

Description

You will design expert-level Terminal-Bench tasks by transforming real supply chain planning work into end-to-end challenges. Each task includes scenario files/data exports, a working solution (e.g., a 5-day production plan solvable in 60–75 minutes), and an automated correctness check. You will document rules/SOPs and edge cases, then iterate based on reviewer feedback to keep instructions unambiguous and results reproducible from the command line.

Responsibilities

• Design expert-level Terminal-Bench tasks from real supply chain tasks so an AI agent can solve end to end
• Define the files, data exports, and systems involved (e.g., ERP, MES, WMS extracts)
• Write a working solution and confirm the task is solvable in 60–75 minutes
• Build automated validation using tests or scripts that accept valid answers and catch planning errors objectively
• Document written rules, constraints, and SOPs, including realistic edge cases, data inconsistencies, and common planner errors
• Iterate tasks based on reviewer feedback; keep data consistent and verification reproducible in a command-line environment

Requirements

• 3+ years of hands-on supply chain or production planning experience (or a PhD with active research in the subfield)
• Experience within the last 3 years; can name the systems, tools, data, and rules used
• Personally executed rule-driven processes (SOPs/planning rules/constraints)
• Worked inside systems of record (ERP, WMS) and used their exports/reports
• Can describe edge cases, data inconsistencies, and common planner errors and how a mistake would be detected
• Technical fluency in Python (or another language) plus command-line; familiarity with files plus CSV/JSON, APIs, or databases
• Available to start around November 4, 2026

Benefits

• Task-based engagement paid per task (each task ~60–75 minutes)
• Client-reviewed and client-approved tasks; vetted through task work rather than a live interview