Research Scientist (Generative & Agentic AI)

TL;DR

Research Scientist (Generative & Agentic AI): Developing and optimizing LLMs, VLMs, and autonomous AI agents that reason and use tools with an accent on post-training techniques like RLHF and SFT. Focus on improving model reasoning, test-time scaling, and bridging frontier research with real-world product impact.

Location: Taipei, Taiwan

Company

hirify.global is a SaaS company that uses artificial intelligence to power business decision-making and turn AI into ROI.

What you will do

• Research and build agentic AI systems focusing on reasoning, planning, tool use, and multi-agent collaboration.
• Advance post-training techniques including SFT, RLHF, and preference optimization to improve model alignment and reliability.
• Optimize the performance, efficiency, and scalability of foundation models across training and inference.
• Design rigorous evaluations and benchmarks for models and agents in real-world scenarios.
• Collaborate with cross-functional teams to ship frontier research into production applications.
• Track AI/ML research trends and publish key findings at leading venues.

Requirements

• Master's degree or Ph.D. in Computer Science, Electrical Engineering, Mathematics, or a related field.
• Deep expertise in at least one of: LLMs, VLMs/multimodal models, Reinforcement Learning, or agentic systems.
• Hands-on experience with fine-tuning, RAG, agent frameworks, or product prototyping.
• Proficiency in Python and PyTorch for building and optimizing models.
• Strong ability to analyze model behavior and diagnose training/inference bottlenecks.

Nice to have

• Publications in top AI/ML conferences such as NeurIPS, ICML, ICLR, CVPR, ACL, or EMNLP.
• Experience with large-scale distributed training or LLM post-training pipelines.
• Contributions to open-source agent frameworks, models, or benchmarks.

Culture & Benefits

• Opportunity to work at the frontier of generative and agentic AI.
• Collaborative and fast-paced team environment.
• Direct impact by bridging high-level research with real-world applications.
• Support for publishing research at top-tier international venues.