Your Impact at LILA

Lila Sciences builds AI systems that accelerate discovery across the physical and life sciences. Our team turns unstructured scientific knowledge, including literature, patents, and technical reports, into structured signals that power downstream Lila applications.

As a Data Extraction Co-Op, you will work alongside research scientists and engineers on a focused sub-problem in this stack. You will fine-tune and evaluate models for scientific knowledge extraction and retrieval, build pipelines for messy real-world data, and ship work that flows into production systems.

What You'll Be Building

• Fine-tune and evaluate language, multimodal, or specialized models for scientific data extraction
• Build retrieval and RAG pipelines surfacing relevant documents, tables, and figures
• Generate and validate structured datasets extracted from scientific literature and patents
• Run end-to-end pipelines and analyze results against clear evaluation criteria
• Document findings and present them to the team
• Contribute to a publication or open-source project where the work supports it

What You'll Need to Succeed

• Pursuing a Bachelor's, Master's, or PhD in CS, chemistry, materials science, physics, biology, or chemical engineering
• Solid foundation in machine learning fundamentals and Python
• Familiarity with NLP or computer vision concepts
• Curiosity about scientific data and willingness to learn quickly in a research setting
• Clear written and verbal communication

Bonus Points For

• Coursework or projects involving multimodal models or document understanding (OCR, table and figure extraction)
• Experience with LLM applications, embeddings, or retrieval systems (chunking, reranking, vector search)
• Experience building evaluation sets or annotation workflows
• Experience working with messy real-world data
• Interest in scientific document parsing