Senior AI Research Engineer - LLM Training & Agent Systems

About the Role

We are looking for a Senior AI Research Engineer with experience in Large Language Models (LLMs), model training and optimization, and AI agent systems. The ideal candidate has a strong foundation in machine learning and practical experience in developing, evaluating, or deploying modern AI systems.

You will work on improving model capabilities, reasoning, tool utilization, and task execution, while translating research ideas and emerging AI techniques into reliable, production-ready solutions.

Key Responsibilities

LLM Training and Optimization

• Develop and optimize LLM training and post-training workflows, including supervised fine-tuning (SFT), reinforcement learning (RL), preference optimization, or related techniques.
• Improve model capabilities such as reasoning, instruction following, tool use, and task completion.
• Explore techniques for model adaptation, self-improvement, and training efficiency.

AI Agent Systems

• Design and implement agentic AI systems supporting planning, information retrieval, tool calling, and multi-step task execution.
• Develop reliable workflows for research, automation, and other complex AI applications.
• Explore multi-agent collaboration, asynchronous execution, and agent orchestration where appropriate.

Model Evaluation and Benchmarking

• Build and maintain evaluation frameworks for assessing language models across reasoning, coding, instruction following, and agent performance.
• Develop benchmarking pipelines and analysis tools to measure model quality, reliability, and efficiency.
• Analyze model behavior and use evaluation results to guide further improvements.

Inference and Performance Optimization

• Improve LLM inference performance, resource utilization, and deployment efficiency.
• Work with modern inference and model-serving frameworks such as vLLM, SGLang, and Hugging Face Transformers.
• Explore relevant optimization techniques, including quantization, distributed inference, and GPU acceleration.

Research and Technical Innovation

• Investigate emerging techniques in foundation models, reasoning, reinforcement learning, agent systems, and multimodal AI.
• Prototype and evaluate new approaches, translating promising research into practical applications.
• Collaborate with research, infrastructure, and product teams to deliver AI capabilities into production.
• Contribute to technical documentation, research reports, publications, or open-source projects where appropriate.

Requirements

• Master's or PhD degree in Computer Science, Artificial Intelligence, Machine Learning, or a related field, or equivalent practical experience.
• Strong programming skills in Python and experience with deep learning frameworks such as PyTorch.
• Practical experience in at least one area of LLM development, such as model training, post-training, evaluation, inference optimization, or agent systems.
• Good understanding of transformer architectures, modern foundation models, and key machine learning concepts.
• Experience developing or applying LLM-based systems, evaluation pipelines, or AI applications.
• Familiarity with model deployment, distributed computing, GPU acceleration, or performance optimization.
• Experience building AI applications involving retrieval, tool use, planning, or multi-step workflows is a plus.
• Ability to independently drive technical projects from experimentation and prototyping to implementation and deployment.

Preferred Qualifications

• Experience working with open-source foundation models such as Qwen, Llama, DeepSeek, or similar models.
• Familiarity with LLM training and optimization techniques, including SFT, RLHF, DPO, GRPO, or related approaches.
• Experience with frameworks and tools such as vLLM, SGLang, Hugging Face Transformers, DeepSpeed, or other distributed training and inference frameworks.
• Knowledge of inference optimization techniques, including quantization, mixed-precision computation, and parallelism strategies.
• Experience with AI agents, multi-agent systems, retrieval-augmented generation (RAG), knowledge graphs, or deep research applications.
• Exposure to multimodal AI, search and recommendation systems, model safety, or adversarial robustness.
• Research publications, open-source contributions, or experience applying research techniques to real-world AI systems.