# ProfiLLM: Utility-Aligned Agentic User Profiling for Industrial Ride-Hailing Dispatch

arXiv: https://arxiv.org/abs/2606.18803 · HTML: https://mlsanity.com/paper/2606.18803/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use.
- Builds on 19 of its 40 references that are in the map.
- First posted on arXiv 17 Jun 2026.

## Placed in

- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)
- LLM alignment / RLHF / preference optimisation / [Process reward models for reasoning](https://mlsanity.com/alignment/process-reward-models-for-reasoning/)

## Builds on (most cited first, 19 of 19)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Toolformer: Language Models Can Teach Themselves to Use Tools](https://mlsanity.com/paper/2302.04761/) (2023)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Self-Attentive Sequential Recommendation](https://mlsanity.com/paper/1808.09781/) (2018)
- [BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations from Transformer](https://mlsanity.com/paper/1904.06690/) (2019)
- [Data-Copilot: Bridging Billions of Data and Humans with Autonomous Workflow](https://mlsanity.com/paper/2306.07209/) (2023)
- [TrafficGPT: Viewing, Processing and Interacting with Traffic Foundation Models](https://mlsanity.com/paper/2309.06719/) (2023)
- [Widespread Flaws in Offline Evaluation of Recommender Systems](https://mlsanity.com/paper/2307.14951/) (2023)
- [Value Function is All You Need: A Unified Learning Framework for Ride Hailing Platforms](https://mlsanity.com/paper/2105.08791/) (2021)
- [SiriusBI: A Comprehensive LLM-Powered Solution for Data Analytics in Business Intelligence](https://mlsanity.com/paper/2411.06102/) (2024)
- [LettinGo: Explore User Profile Generation for Recommendation System](https://mlsanity.com/paper/2506.18309/) (2025)
- [Limiting Bias from Test-Control Interference in Online Marketplace Experiments](https://mlsanity.com/paper/2004.12162/) (2020)
- [An End-to-End Reinforcement Learning Based Approach for Micro-View Order-Dispatching in Ride-Hailing](https://mlsanity.com/paper/2408.10479/) (2024)
- [DiMA: An LLM-Powered Ride-Hailing Assistant at DiDi](https://mlsanity.com/paper/2503.04768/) (2025)
- [DBAIOps: A Reasoning LLM-Enhanced Database Operation and Maintenance System using Knowledge Graphs](https://mlsanity.com/paper/2508.01136/) (2025)
- [LEADRE: Multi-Faceted Knowledge Enhanced LLM Empowered Display Advertisement Recommender System](https://mlsanity.com/paper/2411.13789/) (2024)
- [Revisiting Task-Oriented Dataset Search in the Era of Large Language Models: Challenges, Benchmark, and Solution](https://mlsanity.com/paper/2512.15363/) (2025)

## Built on by (most cited first, 1 of 1)

- [RideSkill: A Hierarchical Algorithm for Generalized Ride Sharing with LLM-Driven Automatic Evolution](https://mlsanity.com/paper/2609.02250/) (2026)

## Abstract (arXiv)

Bringing Large Language Models (LLMs) into industrial ride-hailing dispatch as semantic feature extractors over platform-scale behavioral logs is a compelling but under-explored data systems problem. Production matching pipelines remain dominated by structured numerical features, yet decisive behavioral signals (e.g., a driver's habitual aversion to certain regions) are inherently contextual and naturally expressible as LLM-generated user profiles. However, scaling such profiling to a live, millisecond-latency dispatcher faces three intertwined constraints rarely addressed together: on a platform with millions of daily orders, logs exceed any LLM's context window by orders of magnitude; most users are long-tail, with too few interactions for per-user profiling; and surface-fluent profiles do not necessarily improve downstream prediction utility. We present ProfiLLM, an agentic LLM data pipeline that operationalizes utility-aligned user profiling for production matching systems through two modules. (1) Tool-Augmented Global Knowledge Mining equips an LLM agent with 27 analytical tools to mine platform-scale data, producing reusable global knowledge, adaptive user clustering rules, and region-level supply-demand priors. (2) Utility-Aligned Profile Exploration generates multiple candidate profiles per cluster, evaluates them via a lightweight downstream utility proxy, iteratively refines the best candidates and constructs preference pairs for DPO fine-tuning. Deployed on DiDi's production dispatcher, ProfiLLM achieves up to +6.14% relative AUC improvement in outcome prediction, up to +4.35% GMV gain in dispatching simulation, and consistent improvements in a 14-day online A/B test including +0.47% GMV, +0.33% Completion Rate, and -0.82% Cancel-Before-Accept rate.