Programming Large Language Models with Algebraic Effect Handlers and the Selection Monad (LMPL 2025)

Who

Shangyin Tan, Guannan Wei, Koushik Sen, Matei Zaharia

Track

LMPL 2025

Time Zone

The program is currently displayed in (GMT+08:00) Perth.

Use conference time zone: (GMT+08:00) PerthSelect other time zone

The GMT offsets shown reflect the offsets at the moment of the conference.

Time Band

By setting a time band, the program will dim events that are outside this time window. This is useful for (virtual) conferences with a continuous program (with repeated sessions).
The time band will also limit the events that are included in the personal iCalendar subscription service.

Display full programSpecify a time band

Save

When

Wed 15 Oct 2025 17:00 - 17:15 at Orchid Small - Neuro-Symbolic Language/Agent Design Chair(s): Yang Feng

Abstract

We present Pangolin, a programming language that treats large language model (LLM) interactions as first-class algebraic effects, enabling systematic composition and control over complex AI systems. By modeling non-deterministic choices of LLM results with selection monads, Pangolin allows programmers to abstract over multiple execution paths and automatically select outcomes based on downstream metrics. Pangolin makes it possible to design and manage LLM-centric pipelines with the clarity and reliability of conventional programming languages.

Shangyin Tan

University of California, Berkeley

United States

Guannan Wei

Tufts University

United States

Koushik Sen

University of California at Berkeley

United States

Matei Zaharia

UC Berkeley

Time Zone

The program is currently displayed in (GMT+08:00) Perth.

Use conference time zone: (GMT+08:00) PerthSelect other time zone

The GMT offsets shown reflect the offsets at the moment of the conference.

Time Band

Display full programSpecify a time band

Save

Session Program

Wed 15 Oct
Displayed time zone: Perth change

16:00 - 17:40	Neuro-Symbolic Language/Agent DesignLMPL at Orchid Small Chair(s): Yang Feng Nanjing University

16:00 15m Talk		Vibe Coding Needs Vibe Reasoning – Improving Vibe Coding with Formal Verificationremote LMPL Jacqueline Mitchell University of Southern California, Yasser Shaaban Workato
16:15 15m Talk		Current Practices for Building LLM-Powered Reasoning Tools Are Ad Hoc—and We Can Do Better LMPL Aaron Bembenek The University of Melbourne Pre-print
16:30 15m Talk		Composable Effect Handling for Programming LLM-integrated Scripts LMPL Di Wang Peking University Pre-print
16:45 15m Talk		The LLM Era Demands Natural-Language-Aligned Theorem Provers for Mathematics LMPL Qinxiang Cao Shanghai Jiao Tong University, Lihan Xie Shanghai Jiao Tong University, Junchi Yan Shanghai Jiao Tong University
17:00 15m Talk		Programming Large Language Models with Algebraic Effect Handlers and the Selection Monad LMPL Shangyin Tan University of California, Berkeley, Guannan Wei Tufts University, Koushik Sen University of California at Berkeley, Matei Zaharia UC Berkeley