LLM Engineer_FedGPT

Taiwan AI Labs is building real-world AI systems powered by large language models, NLP, retrieval, domain knowledge, and tool-using workflows.
We are looking for an LLM Engineer with a strong foundation in machine learning, NLP, model training, evaluation, and practical deployment. This role is suitable for candidates who have hands-on experience with NLP or LLM-related projects, and who want to work on practical AI systems beyond simple demos.
You will work closely with senior engineers and product teams to improve model behavior from experimentation to production.

【Responsibilities】
1. LLM & NLP Model Training
-Prepare datasets for LLM and NLP development, including instruction data and domain-specific text data
-Fine-tune or adapt language models using supervised fine-tuning, preference optimization, or related methods
-Work on NLP tasks such as text classification, named entity recognition, information extraction, semantic matching, and document understanding
-Analyze training results and model outputs to improve model behavior
2. Evaluation & Benchmarking
-Build evaluation datasets, metrics, and benchmark workflows for LLM, NLP, RAG, and agent-related tasks
-Conduct error analysis to identify failures caused by data quality, retrieval, prompting, model reasoning, or generation
-Track model quality across datasets, prompts, model versions, and system changes
3. Applied LLM Systems
-Support RAG and agent workflows, including retrieval, context construction, tool use, and multi-step reasoning
-Assist in model integration, batch inference, API testing, deployment validation, and model behavior monitoring

【Essential Qualifications】
1.Strong foundation in machine learning, NLP, and deep learning fundamentals
2.Hands-on experience fine-tuning or adapting language models using methods such as supervised fine-tuning, instruction tuning, or similar approaches
3.Hands-on experience with NLP tasks such as text classification, NER, relation extraction, document parsing, semantic search, or related information extraction tasks
4.Experience with dataset construction, training / evaluation workflows, and error analysis for NLP or LLM tasks
5.Proficiency in Python and familiarity with ML frameworks
6.Understanding of modern language models, including their training, adaptation, evaluation, and practical limitations
7.Ability to analyze model outputs, diagnose failure patterns, and propose effective improvement strategies
8.Strong ownership, learning ability, and communication skills

【Preferred Qualifications】
1.Experience with preference optimization, reinforcement learning, or other advanced model optimization methods
2.Experience designing evaluation datasets, metrics, or benchmark workflows for LLM-based systems
3.Experience with applied LLM systems such as RAG or agentic workflows, including retrieval, tool use, planning, or multi-step reasoning
4.Understanding of LLM inference and deployment constraints, such as latency, throughput, batching, GPU memory, concurrency, and cost
5.Familiarity with Docker, Linux, APIs, Kubernetes, or cloud environments
6.Experience deploying ML models or integrating models into production systems is a plus

【Why Join Us】
1.Work on real-world LLM systems used in practical enterprise scenarios
2.Improve model behavior through data, training, evaluation, and applied system design
3.Build AI capabilities that move from experimentation to production impact