It's fun to work in a company where people truly BELIEVE in what they're doing!
We're committed to bringing passion and customer focus to the business.
Kyriba is a global leader in liquidity performance that empowers CFOs, Treasurers and IT leaders to connect, protect, forecast and optimize their liquidity. As a secure and scalable SaaS solution, Kyriba brings intelligence and financial automation that enables companies and banks of all sizes to improve their financial performance and increase operational efficiency. Kyriba’s real-time data and AI-empowered tools empower its 3,000 customers worldwide to quantify exposures, project cash and liquidity, and take action to protect balance sheets, income statements and cash flows. Kyriba manages more than 3.5 billion bank transactions and $15 trillion in payments annually and gives customers complete visibility and actionability, so they can optimize and fully harness liquidity across the enterprise and outperform their business strategy. For more information, visit www.kyriba.com.
We are working on innovative approaches to temporal logic processing in Large Language Models (LLMs), specifically through our Temporal Logic Enhancement System (TLES). This system shows promising results in handling complex temporal expressions, and we're looking to develop this work into a comprehensive academic publication and to prepare its industrialization for the development of Kyriba's future agentic framework.
The intern will contribute to developing a rigorous evaluation framework for TLES and expanding its theoretical foundations. The project involves both analytical and practical components, with potential for significant academic impact.
Main Responsibilities:
Design and implement a comprehensive evaluation process for TLES
Develop diverse test cases to assess temporal reasoning capabilities
Conduct comparative analysis with existing models (both proprietary and open-source), and conduct a state-of-the-art review on the subject.
Document theoretical frameworks around TLES's capabilities and limitations
Contribute to academic paper preparation
Technical Aspects:
Benchmark TLES against various other techniques using open-source models (including but not limited to variously sized Llama v3 models) available through Databricks as well as other closed source ones like OpenAi’s and Anthropics’.
Analyze and document the system's coverage of temporal expressions
Optional: Develop an agent-based approach for flexible temporal composition
Required Skills:
Master's student in Computer Science, Data Science, or related field
Strong programming skills (Python preferred)
Knowledge of NLP and Large Language Models
Familiarity with academic research methodologies
Excellent analytical and documentation skills
Fluency in English
Desired Skills:
Experience with LLM evaluation frameworks
Research appetite
Familiarity with Databricks
