AI Model Engineer

Lenovo is seeking a highly motivated Foundation Model Engineer to contribute to the design, development, and exploration of our next-generation AI systems. As a Foundation Model Engineer, you will focus on adapting and improving foundation models for real products: pre-training, fine-tuning, post-training (RLHF/DPO), and evaluation. You'll work closely with research and platform teams to turn large models into reliable, high-quality systems. This is an exciting opportunity to gain hands-on experience with cutting-edge AI systems while collaborating with experienced engineers, researchers, and product teams to help advance Lenovo's Hybrid AI vision and make Smarter Technology for All.

Key Responsibilities

• Fine-tune and adapt foundation models
  Design and run fine-tuning and parameter-efficient training (e.g., LoRA, adapters, low-rank methods) for LLMs and related models to support specific products and domains.

• Implement post-training pipelines
  Build and maintain pipelines for instruction tuning, preference optimization (e.g., RLHF, DPO), and tool-use training to improve helpfulness, safety, and controllability.

• Data curation and labeling strategy
  Work on dataset creation, filtering, deduplication, and labeling strategies; collaborate with data and annotation teams to define high-quality training and evaluation sets.

• Optimize training performance
  Profile and optimize training jobs for throughput and cost efficiency across GPUs/accelerators (batching, sharding, mixed precision, memory optimization).

• Model evaluation & benchmarking
  Design and run evaluation suites (automatic metrics + human evals), analyze regressions, and compare model variants across internal and external benchmarks.

• Production readiness
  Partner with infra and product teams to move trained checkpoints into production, including versioning, rollout strategies, and monitoring model behavior in the wild.

• Stay current with the field
  Track advances in model architectures, optimization methods, and training techniques; propose and run experiments to bring relevant ideas into the stack.

Qualifications

• 2+ years of industry experience in ML, applied research, or highly relevant internships/research roles.

• Master's degree or PhD in Computer Science, Computer Engineering, Electrical Engineering, or a related technical field.

• Strong programming skills in Python.

• Deep familiarity with PyTorch, TensorFlow, or JAX.

• Solid understanding of deep learning fundamentals: optimization, regularization, initialization, distributed training basics.

• Hands-on experience training or fine-tuning large models (not just calling APIs): e.g., language models, vision-language models, or similar.

• Comfort working with training pipelines, experiment tracking, and model checkpoints.

• Experience building datasets for training/eval and running structured experiments.

• Ability to analyze results, understand failures, and iterate quickly.

Bonus Points:

• Experience with LLM-specific training techniques (instruction tuning, tool-use tuning, chat models).

• Experience with RLHF or preference-based learning (e.g., reward models, DPO, PPO-style training).

• Familiarity with distributed training frameworks (e.g., DeepSpeed, FSDP, Megatron, XLA, custom sharding strategies).

• Prior work on safety, bias, robustness, or alignment of large models.

• Publications, open-source contributions, or strong public projects in large-scale ML.

#LATC