Mathematical Reasoning AI Systems Engineer

OpenTrain AI, Inc.; United States; Remote; USD 83,000 - 131,000; Full time

Benefits offered by this job

Remote work worldwide

Four-week contractor engagement

Overlap with Pacific Time

Job summary

OpenTrain AI, Inc. seeks a contractor to turn mathematical proofs and dense academic material into structured software systems, automated pipelines, and reusable tools for AI agents.

You will build multi-step agent workflows for reasoning and validation, translating proofs into technical specs and pipelines. You will work with researchers to create modular, reliable software and evaluate execution traces to improve systems and outputs.

Qualifications

• Advanced Python proficiency and practical experience with LLM orchestration.
• Experience designing multi-step agent workflows for complex reasoning.
• Ability to translate math/CS concepts into structured software systems.

Responsibilities

• Design, build, and optimize multi-step agent workflows for multi-hop reasoning, hypothesis generation, and technical validation.
• Translate mathematical proofs and academic material into technical specs and automated pipelines.
• Develop architecture frameworks, tools, and memory systems for STEM-focused AI agents.
• Evaluate agent performance for accuracy, reliability, and traceability across complex execution steps.
• Collaborate with mathematical researchers and subject-matter experts on reusable AI workflows.

Skills

Python proficiency

Large language model orchestration

Multi-hop reasoning

Agent workflows design

Software engineering

Tools

LangGraph

AutoGen

CrewAI

Job description

The work

You will turn mathematical proofs and dense academic material into structured software systems, automated pipelines, and reusable tools for AI agents. The work combines mathematical reasoning with hands-on engineering and system evaluation.

You will work with mathematical researchers and other STEM experts to turn abstract ideas into modular, reliable software. You will also review agent outputs and execution traces, then improve the systems based on what you find.

• Design, build, and optimize multi-step agent workflows for multi-hop reasoning, hypothesis generation, and technical validation.
• Translate mathematical proofs and other complex academic material into technical specifications and automated pipelines.
• Develop architecture frameworks, tools, and memory systems for STEM-focused AI agents.
• Evaluate agent performance for accuracy, reliability, and traceability across complex execution steps.
• Collaborate with mathematical researchers and subject-matter experts on reusable AI workflows.

What it pays and takes

The listing does not provide a pay rate. This is a remote, four-week contractor engagement with a 40-hour weekly commitment and four hours of overlap with Pacific Time.

• Remote role open worldwide.
• Four-week contractor engagement.
• 40 hours per week, with four hours of overlap with Pacific Time.
• English-language work.
• Production experience architecting and deploying multi-step agentic systems.
• Advanced Python proficiency and practical experience with large language model orchestration.
• Ability to generate structured outputs, integrate tools, and manage application state.
• Experience building AI or machine learning applications in mathematics, physics, computer science, or quantitative finance.
• Helpful experience includes LangGraph, AutoGen, CrewAI, or custom agent frameworks.
• Relevant background may include multi-hop reasoning, technical validation, agent memory systems, traceable execution evaluation, or work with STEM researchers.

About AI training work

AI training work is the human side of building AI systems. Engineers and subject-matter experts help create, test, and evaluate model behavior, and specialized knowledge matters when systems must handle rigorous mathematical or scientific tasks.