Team

AMAP-ML Research Team

I lead the AMAP-ML team at Alibaba Group, a 100+ member research team focused on building the next generation of intelligent systems for Alibaba’s mapping and mobility platform.

Our mission: Advance the frontiers of AI research while building systems that directly improve the daily lives of hundreds of millions of users. We believe the strongest research comes from the intersection of academic rigor and real-world impact.

Research Directions

LLM

LLM Reasoning & Agents

Reinforcement learning for model reasoning, tree-search agent training, agent-data co-evolution, and autonomous driving VLA models.

GPG · Tree-GRPO · CoEvolve · MathForge · AutoDrive-R

Generation

Image & Video Generation

Diffusion model optimization, video generation and understanding, virtual try-on, 3D editing, and generation quality evaluation.

DCW · S-Guidance · Eevee · NarrLV · VMBench · FingER

Architecture

Foundation Architectures

Novel attention mechanisms, unified pretraining for generation and understanding, end-to-end pixel generation, and diffusion language models.

FASA · USP · EPG · SCALAR · AR-MAP

Multimodal

Multimodal Understanding

Vision-language reasoning, visual policy optimization, spatial intelligence evaluation, and multimodal in-context learning.

MobileVLM · VisionLLaMA · SpatialGenEval · UniVG-R1

Mobility

Intelligent Mobility

Route-planning agent benchmarks, generative navigation, map-augmented reasoning, and integrated search-recommendation for travel.

MobilityBench · GenMRP · IntTravel · DSFNet

Perception

Detection & 3D Perception

Object detection, semantic segmentation, 3D perception, and open-vocabulary detection for real-world applications.

YOLOv6 · UPRE · PLUG · SCTNet · AeDet

Open Source

Our work is available on GitHub at AMAP-ML, with 20+ active projects and 10,000+ cumulative stars across all organizations we’ve contributed to.

Project	Stars	Description
YOLOv6	5,700+	Industrial-grade real-time object detection
MobileVLM	1,500+	Mobile vision language model
Twins	1,000+	Spatial attention Vision Transformer
GPG	—	RL reasoning baseline (ICLR 2026)
USP	—	Unified self-supervised pretraining (ICCV 2025)
Tree-GRPO	—	Tree search for LLM agent RL (ICLR 2026)

Join Us

We are always looking for talented interns and full-time researchers to join our team. We offer:

What We Offer

Competitive compensation and benefits
Publish at top venues (CVPR, NeurIPS, ICLR, ACL, etc.)
Work with state-of-the-art GPU clusters
Collaborate with researchers from top global AI labs
Open-source culture: your work will be seen by the community

What We Look For

Strong background in ML/CV/NLP/RL
Publications at top venues (preferred but not required for interns)
Solid coding skills and engineering mindset
Self-motivated with strong research taste
Passion for building AI systems that ship

Interested? Please send your CV and representative publications to cxxgtxy@gmail.com. We review all applications and respond promptly.

Xiangxiang Chu（初祥祥）