Role Overview

We are searching for ambitious, boundary-pushing, and technically exceptional Class of 2027 graduates (open to Bachelor's, Master's, and Ph.D. candidates) to join our AI Research Labs.

In this role, you will not be a cog in a machine. You will directly contribute to the next generation of Artificial General Intelligence (AGI) and Embodied AI. We care less about your years on paper and more about the elegance of your code, the sharpness of your technical intuition, and your drive to solve open-ended problems.

💡 Core Technical Tracks (Deep dive into one or bridge multiple)

1. Large Language Models & Multimodal AI

• Pre-training: Design and optimize next-generation model architectures, spearhead massive-scale distributed training stability, and optimize compute/infrastructure efficiency.
• Post-training: Advance SFT, RLHF, and RLAIF (PPO, DPO, etc.) methodologies to push the limits of model reasoning, math, coding, and long-horizon planning.
• Data Synthesis & Engineering: Develop advanced algorithms for automated data curation, high-fidelity synthetic data generation, and scalable cleaning pipelines to build data-driven flywheels.
• Evaluation: Construct rigorous, adversarial evaluation frameworks and automated evaluation agents to quantifiably diagnose model capabilities and guide algorithmic iterations.

2. Generative Models & Next-Gen Architecture

• Diffusion & World Models: Explore text-to-image/video dynamics, reasoning-heavy "slow-thinking" architectures, and physical-world-grounded World Models to build predictive AI systems.

3. Agentic AI

• Agent Systems: Architect sophisticated agent frameworks equipped with long-term memory, autonomous tool use, reflection, and multi-agent collaboration to transition AI from content generation to autonomous execution.

4. Embodied AI & Robotics

• Embodied Perception & Control: Develop Vision-Language-Action (VLA) models to achieve seamless closed-loop perception, decision-making, and control.
• Dexterous Manipulation: Focus on tactile sensing integration, reinforcement learning for high-degree-of-freedom control, and cross-modal coordination to enable high-precision, compliant robotic manipulation.

🎯 Who We Are Looking For

• Target Group: Class of 2027 graduates (Graduation date between September 2026 and August 2027) majoring in Computer Science, Automation, Robotics, Mathematics, Electrical Engineering, or related quantitative fields.
• Solid Engineering Fundamentals: Exceptional coding skills with deep proficiency in PyTorch or JAX. Familiarity with leveraging AI-assisted development tools (e.g., Cursor, open-source agent frameworks) to maximize workflow automation is a strong plus.
• Standout Qualities (Meeting any of the following makes you a priority candidate):

1. Top-Tier Publications: First-author (or core contributor) papers at premier AI conferences (e.g., ICLR, NeurIPS, CVPR, ICCV, ACL, ICRA, IROS).
2. Competitive Edge: Top tier placements in global hackathons, Kaggle, ACM-ICPC, or prestigious robotics competitions (e.g., RoboCup).
3. Open-Source Contributors: Active contributors or maintainers of impactful open-source AI repositories, LLM tooling, or agent frameworks.
4. Hands-on Lab Experience: Prior deep immersion or internship experience at leading AI labs or industry research groups, working on large-scale training, RL alignment, or physical hardware deployment.

🌟 What We Offer

• Uncapped Compute Power: Access to state-of-the-art, massive GPU clusters. Your ideas will never be bottlenecked by infrastructure.
• Fluid Cross-Disciplinary Growth: We break down organizational silos. You are encouraged to collaborate and rotate between LLM, Agentic, and Embodied AI teams.
• Elite Mentorship: 1-on-1 mentorship from industry-recognized experts and top-tier conference reviewers, helping you navigate the full lifecycle from lab-scale research to production-grade deployment.
• Highly Competitive Compensation: Tailored SSP / Top Talent offer packages with uncapped compensation for exceptional individuals.