VirtualMLE: A Virtual ML Engineer that Optimizes Sequential Recommenders

2026-06-02

AI Summary

This work addresses the heavy reliance on manual trial-and-error in tuning sequential recommendation models on new datasets, a process that lacks automation. To overcome this limitation, the authors propose the first large language model (LLM)-based agent framework that enables automatic hyperparameter and architectural tuning through a closed-loop pipeline of execution, reflection, and hierarchical memory updating. The framework incorporates a novel cognitive summary distillation technique to facilitate heuristic knowledge transfer across datasets and is integrated with backbone models such as SASRec and HSTU. Experimental results on three Amazon benchmark datasets demonstrate that the proposed method achieves competitive performance with significantly fewer trials, thereby substantially accelerating the tuning process.

Abstract

Recent advancements in Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning, reflection, and tool utilization, unlocking new paradigms for automating complex engineering workflows. However, in the domain of sequential recommendation (SR), tuning models on new datasets still relies heavily on the manual trial-and-error of experienced machine learning engineers. To bridge this gap, we propose \textbf{VirtualMLE}, an LLM-agent framework that leverages the cognitive capabilities of LLMs to organize recommender optimizing into a closed loop of execution, reflection, and memory update. After each trial, the agent explicitly analyzes the observed outcomes and stores concise heuristic feedback in a hierarchical memory system. We evaluate VirtualMLE on three Amazon SR benchmarks with two representative backbones, SASRec and HSTU. VirtualMLE reaches competitive recommendation quality with substantially fewer trials. Furthermore, we observe that cognition summaries distilled from previous datasets can significantly accelerate the search process on unseen datasets, demonstrating the potential of transferring tuning heuristics. Overall, our results provide compelling evidence that LLM agents equipped with reflection and memory can serve as practical virtual engineers to automate and amortize heuristic learning in SR optimization. Our codes are available.

Problem

• sequential recommendation
• model tuning
• automated optimization
• machine learning engineering
• heuristic learning

Innovation

• LLM agent
• sequential recommendation
• automated optimization
• reflection and memory
• heuristic transfer