Kyriba empowers CFOs, Treasurers, and their IT counterparts to transform treasury, payments, working capital, and connectivity solutions to activate liquidity as a dynamic, real-time vehicle for growth and value creation. Kyriba is a secure, scalable SaaS platform that leverages artificial intelligence, automates payments workflows, and enables thousands of multinational corporations and banks to maximize growth, protect against loss from fraud and financial risk, and reduce operational costs. With 2,000 clients worldwide, including 25% of Fortune 500 and Eurostoxx 50 companies, Kyriba manages more than 1.3 billion bank transactions per year, and 250 million payments for a total value of $15 Trillion annually. Kyriba is headquartered in San Diego, with offices in Dubai, Frankfurt, Londo, Paris, Shanghai, Singapore, Tokyo, Warsaw and other major locations. For more information, visit www.kyriba.com.

Role: 

Large Language Models (LLMs) are increasingly integral to data science workflows. Present practices, largely rudimentary, involve manually optimizing prompts. For production-level applications, a more scientific methodology is essential. This internship aims to develop strategies for optimizing pipelines employing LLMs. Emerging frameworks like DSPY and Llamaindex are gaining attention. The primary goal is to implement a pipeline dedicated to the optimization and compilation of LLMs.

Profile:

Student engineering school or master's degree in data science, statistics, mathematics or applied mathematics

Transformer architecture

Convex and non-convex optimization

Solid knowledge of Python

Effective communication skills in English

Ability to work in a team

Curiosity