V7 is looking for an AI Engineer to lead the development and evaluation of synthetic data pipelines used to train frontier AI models. You'll design modular, reproducible systems and collaborate closely with researchers and ML practitioners.

What You'll Do

Design, implement, and maintain synthetic data generation pipelines for multi-modal training tasks.
Evaluate pipeline output using well-grounded proxy metrics and sound statistical experiments.
Own the design and execution of experiments involving LLMs, ensuring high reproducibility and clarity of findings.
Apply agentic design patterns and context engineering techniques to maximize model performance.
Use tools like Cursor, GitHub Copilot, and LLM agents to accelerate iteration, debugging, and documentation.
Collaborate with researchers and engineers across the stack to translate experimental insights into scalable systems.

What We're Looking For

3+ years of software engineering experience with at least one major programming language (Python or JavaScript preferred).
A strong academic background with an MS or higher in Computer Science, Engineering, Mathematics, or a related scientific field.
Deep familiarity with Git, DVC, shell environments, and data pipeline orchestration.
A solid foundation in statistics and experimental design, especially in the context of ML evaluation.
Experience working with LLM systems, including prompt and context engineering, agentic workflows, and output optimization and reliability strategies.

Nice to Have

Familiarity with recent research on LLM training datasets and evaluation benchmarks, including CoDA, ChartGalaxy, Chain of Functions, and ChartQA-X.

Technical Stack

Python
JavaScript
Git
DVC
LLM systems

Team & Environment

You'll join a high-impact team at the forefront of AI research and engineering. We value curiosity, a bias toward iteration and improvement, and the ability to thrive in fast-moving environments without clearly defined playbooks. We prefer modular, reproducible systems over ad-hoc experimentation and apply rigour in both code and evaluation.

V7 champions equality and inclusion because diverse teams build better products.