We are looking for talented individuals to join our team in 2027. As a graduate, you will get opportunities to pursue bold ideas, tackle complex challenges, and unlock limitless growth. Launch your career where inspiration is infinite at our Company.
Successful candidates must be able to commit to an onboarding date by end of year 2027. Please state your availability and graduation date clearly in your resume.
Team Introduction:
The Applied Machine Learning Ark team combines system engineering and machine learning to develop and operate Large Language Model (LLM) service platforms that offer businesses Model-as-a-Service (MaaS) solutions, serving both large model providers and downstream users. The US team drives the design, development, and operation of MaaS solutions across the US and international markets outside mainland China. We are building full-stack, end-to-end solutions spanning text and multimodal LLM algorithms, LLM training/fine-tuning/inference frameworks, prompt engineering, model alignment, and intelligent agent systems. Beyond model serving, we operate large-scale log analytics pipelines that process massive volumes of invocation logs from text models, multimodal models, and agent systems — extracting usage patterns, quality signals, and actionable insights to inform model improvement, system optimization, and product decisions through continuous, data-driven feedback loops. We are actively seeking talented engineers and researchers specializing in Large Language Models and AI Agent systems to join our dynamic team.
Topic Content:
As model capabilities improve and computation becomes cheaper, the key challenge in real-world deployment is no longer building a capable one-off assistant, but building agent systems that improve through use. This research studies a self-evolving agent framework in which execution traces, environmental responses, and human feedback are converted into signals for continual improvement. The goal is to establish a closed loop from execution to feedback, attribution, accumulation, and reuse, so that system capability grows with real-world interaction. We focus on three tightly coupled directions: adaptive runtime, which enables online adjustment of planning, tool use, and control policies; experience compilation, which abstracts reusable skills, rules, and failure patterns from trajectories; and evaluation-governance loops, which ensure that each system update is measurable, comparable, and reversible. Together, these components support a synergistic co-evolution of the model layer and the harness layer, improving task quality, reducing manual intervention, and accumulating durable capability over time. More broadly, this work reframes agent deployment as a continual learning systems problem: not how to build a stronger static agent, but how to build an operational system that learns reliably from experience.
Responsibilities:
- Research and develop agent frameworks that continuously learn and improve from execution traces, user feedback, and environmental signals.
- Build large-scale log analytics pipelines to extract quality signals, usage patterns, and actionable insights from model and agent invocation logs, driving data-informed system and model improvements.
- Explore and apply frontier techniques in LLM post-training, reasoning, and planning to enhance agent capabilities.
- Collaborate across algorithm research, platform engineering, and product teams to turn research ideas into production-grade systems at scale.
Minimum Qualifications:
- Individuals who are completing or have recently completed a Ph.D. in Computer Science, Artificial Intelligence, Machine Learning, or a closely related discipline.
- Strong theoretical and practical foundation in machine learning, deep learning, reinforcement learning, or optimization.
- Research experience in at least one of the following areas: LLM-based agents, planning and reasoning, multi-agent systems, continual/lifelong learning, or LLM post-training (e.g., RLHF, DPO, GRPO, self-play).
- Strong programming skills in Python and proficiency with ML frameworks (e.g., PyTorch, TensorFlow, JAX).
- Publication record at top-tier venues (e.g., NeurIPS, ICML, ICLR, ACL, EMNLP, NAACL, AAAI, AAMAS, COLM).
- Strong problem-solving skills and ability to thrive in a fast-paced, collaborative environment.
Preferred Qualifications:
- Publications in areas directly related to agent learning and adaptation, such as tool use, self-improvement, skill discovery, trajectory optimization, reward modeling, or agent evaluation.
- Research experience in LLM reasoning and planning, including chain-of-thought, tree/graph search, Monte Carlo methods, or inference-time compute scaling.
- Experience training or fine-tuning large language models, including supervised fine-tuning, preference optimization, or curriculum learning.
- Hands-on experience building or evaluating LLM-based agent systems (e.g., ReAct, function calling, code generation agents, or multi-agent orchestration).
- Familiarity with meta-learning, few-shot generalization, or transfer learning in the context of LLM-based systems.
- Experience with feedback-driven optimization loops, such as online learning, bandit methods, or evolutionary strategies applied to agent improvement.
- Strong interest in bridging frontier AI research with production-grade engineering — turning papers into systems that work at scale.
- Internship experience at technology companies or research organizations.
