# Learning to Staff: Offline Reinforcement Learning and Fine-Tuned LLMs for Warehouse Staffing Optimization

Source: [Learning to Staff: Offline Reinforcement Learning and Fine-Tuned LLMs for Warehouse Staffing Optimization](<https://www.sota2.com/research/paper/69c60ee90636e5bf4afc83b6>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-28T05:00:04.914Z

Authors: Kalle Kujanp\\"a\\"a, Yuying Zhu, Kristina Klinkner, Shervin Malmasi

Published: 2026-03-25T23:56:25.000Z

## Abstract

We investigate machine learning approaches for optimizing real-time staffing decisions in semi-automated warehouse sortation systems. Operational decision-making can be supported at different levels of abstraction, with different trade-offs. We evaluate two approaches, each in a matching simulation environment. First, we train custom Transformer-based policies using offline reinforcement learning on detailed historical state representations, achieving a 2.4% throughput improvement over historical baselines in learned simulators. In high-volume warehouse operations, improvements of this size translate to significant savings. Second, we explore LLMs operating on abstracted, human-readable state descriptions. These are a natural fit for decisions that warehouse managers make using high-level operational summaries. We systematically compare prompting techniques, automatic prompt optimization, and fine-tuning strategies. While prompting alone proves insufficient, supervised fine-tuning combined with Direct Preference Optimization on simulator-generated preferences achieves performance that matches or slightly exceeds historical baselines in a hand-crafted simulator. Our findings demonstrate that both approaches offer viable paths toward AI-assisted operational decision-making. Offline RL excels with task-specific architectures. LLMs support human-readable inputs and can be combined with an iterative feedback loop that can incorporate manager preferences.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Operational decision-making on hand-crafted simulator](<https://www.sota2.com/research/sota/operational-decision-making-on-hand-crafted-simulator.md>): Operational decision-making; hand-crafted simulator; Throughput Improvement: 0.6.

- [Operational Throughput Optimization on Internal Learned Simulators (held-out shifts)](<https://www.sota2.com/research/sota/operational-throughput-optimization-on-internal-learned-simulators-held-out-shifts.md>): Operational Throughput Optimization; Internal Learned Simulators (held-out shifts); Throughput Improvement (%): 2.4.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2603.24883>)

[PDF](<https://arxiv.org/pdf/2603.24883.pdf>)
