Goodlight AI is building intelligent agents that automate complex workflows across retail, data, and enterprise systems. We’re looking for a Machine Learning Intern who is excited about applied AI—someone who wants to go beyond models and help ship real, production-grade agent systems.

This role is ideal for someone who enjoys working across the stack: from data pipelines and model experimentation to integrating LLMs into live products.

What you’ll do

• Build and experiment with ML models and LLM-powered systems (classification, embeddings, retrieval, agents).

• Work on real-world use cases such as retail personalization, workflow automation, and customer intelligence.

• Design and improve prompt pipelines, evaluation frameworks, and agent behaviors.

• Collaborate on data pipelines: ingestion, cleaning, feature engineering, and analysis.

• Prototype and ship features quickly in a production environment.

• Contribute to internal tooling for model monitoring, evaluation, and iteration.

What we’re looking for

• Strong fundamentals in machine learning and statistics.

• Hands-on experience with Python and ML libraries (e.g., PyTorch, scikit-learn).

• Familiarity with LLMs, embeddings, vector databases, or agent frameworks is a strong plus.

• Comfort working with APIs and integrating ML into applications.

• Ability to move fast, experiment, and learn independently.

• Clear thinking and ability to translate messy problems into structured solutions.

Bonus points

• Experience building AI agents or working with tools like LangChain, OpenAI APIs, etc.

• Exposure to retail, ecommerce, or customer data problems.

• Experience with data engineering (SQL, pipelines, ETL).

• Side projects or demos showcasing applied ML work.

What you’ll get

• Direct exposure to building cutting-edge AI agents and production systems.

• Opportunity to work closely with the founding team on high-impact problems.

• Ownership of real features that go live to users.

• Fast learning environment with high autonomy.

Logistics

• Location: Remote

• Duration: 3–6 months

• Potential for full-time conversion based on performance