Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I’m Hao jie Li. I’m an AI Engineer with a Master of Science in Information and Communication Technology from Chalmers University, focused on production-ready Large Language Models, deep learning systems, and multi-agent AI. I build end-to-end AI systems including RAG pipelines, knowledge-graph integration, and distributed training on HPC clusters, with a strong focus on reliability and scalability. My work spans 6G wireless AI and trustworthy LLM research, emphasizing rigorous model alignment via reinforcement learning and careful data curation. I enjoy turning complex research into robust, production-ready tools that solve real-world problems and push the boundaries of what’s possible with AI.…Hi, I’m Hao jie Li. I’m an AI Engineer with a Master of Science in Information and Communication Technology from Chalmers University, focused on production-ready Large Language Models, deep learning systems, and multi-agent AI. I build end-to-end AI systems including RAG pipelines, knowledge-graph integration, and distributed training on HPC clusters, with a strong focus on reliability and scalability. My work spans 6G wireless AI and trustworthy LLM research, emphasizing rigorous model alignment via reinforcement learning and careful data curation. I enjoy turning complex research into robust, production-ready tools that solve real-world problems and push the boundaries of what’s possible with AI.

Hao jie Li

AI Engineer, Data Scientist, Full Stack Developer, +2





Hi, I’m Hao jie Li. I’m an AI Engineer with a Master of Science in Information and Communication Technology from Chalmers University, focused on production-ready Large Language Models, deep learning systems, and multi-agent AI. I build end-to-end AI systems including RAG pipelines, knowledge-graph integration, and distributed training on HPC clusters, with a strong focus on reliability and scalability. My work spans 6G wireless AI and trustworthy LLM research, emphasizing rigorous model alignment via reinforcement learning and careful data curation. I enjoy turning complex research into robust, production-ready tools that solve real-world problems and push the boundaries of what’s possible with AI.…Hi, I’m Hao jie Li. I’m an AI Engineer with a Master of Science in Information and Communication Technology from Chalmers University, focused on production-ready Large Language Models, deep learning systems, and multi-agent AI. I build end-to-end AI systems including RAG pipelines, knowledge-graph integration, and distributed training on HPC clusters, with a strong focus on reliability and scalability. My work spans 6G wireless AI and trustworthy LLM research, emphasizing rigorous model alignment via reinforcement learning and careful data curation. I enjoy turning complex research into robust, production-ready tools that solve real-world problems and push the boundaries of what’s possible with AI.

Available to hire

Hi, I’m Hao jie Li. I’m an AI Engineer with a Master of Science in Information and Communication Technology from Chalmers University, focused on production-ready Large Language Models, deep learning systems, and multi-agent AI. I build end-to-end AI systems including RAG pipelines, knowledge-graph integration, and distributed training on HPC clusters, with a strong focus on reliability and scalability.

My work spans 6G wireless AI and trustworthy LLM research, emphasizing rigorous model alignment via reinforcement learning and careful data curation. I enjoy turning complex research into robust, production-ready tools that solve real-world problems and push the boundaries of what’s possible with AI.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

Chinese

Fluent

English

Fluent

Swedish

Intermediate

Work Experience

Algorithm Intern – In Context Learning for 6G MIMO at Huawei Sweden Research Center, Shannon Research Center, Wireless Network Algorithm Lab

October 1, 2025 - Present

Architected a novel non-causal Transformer foundation model with SwiGLU activations, RMSNorm, and FlashAttention to replace traditional channel estimation pipelines in 6G TDD Massive MIMO systems, processing 64-antenna real-time signals with implicit beamforming via 1x1 convolutional feature projection. Pioneered hybrid optimization combining Muon optimizer for 2D parameters and a Warmup-Stable-Decay scheduler, achieving 3.2x faster convergence versus AdamW on wireless channel datasets. Deployed distributed training across 5 GPUs using DeepSpeed, achieving 4.8x speedup with near-linear scaling efficiency, enabling rapid iteration on 19.9M-parameter models trained on 500K channel snapshots. Impact: Reduced Symbol Error Rate (SER) by 60% in 64-antenna scenarios (0.012→0.0048), narrowing the gap to theoretical upper bound (Perfect CE + LMMSE) to ≤1dB at SER=0.1, demonstrating production-ready performance for next-generation wireless systems. Established rigorous benchmarking framework

Research Assistant – Trustworthy AI at Chalmers University of Technology

January 1, 2025 - July 1, 2025

Designed and released a comprehensive Chinese textual ambiguity benchmark comprising 900 context-disambiguation pairs across 3 linguistic families and 9 subtypes, with an open-source evaluation harness enabling reproducible assessment of LLM reliability under linguistic uncertainty. Benchmarked state-of-the-art models (Gemma-2, Qwen-3, DeepSeek-R1) across six prompting strategies, achieving Macro-F1 87.0 in ambiguity detection and Set-F1 67.7 in interpretation. Developed lightweight RoBERTa-based ambiguity detector (94.7% accuracy) and conducted systematic over-confidence analysis using an LLM-as-judge framework with NLI-based contradiction checks, uncovering critical reliability gaps in production LLM deployments.