postdoc, research in LLM engineering i.e.

EPFL; Lausanne; Vor Ort; EUR 90’000 - 120’000; Vollzeit

Zusammenfassung

The EPFL is seeking a technically exceptional Postdoctoral Researcher in LLM engineering to lead the MediHealth initiative. This role involves design of scalable LLM pipelines, model development, safety alignment, and collaboration with researchers for advancements in trustworthy AI for medical applications. Candidates should have substantial experience in training models and a strong skillset in PyTorch.

Leistungen

Competitive salary

Collaborative team culture

Opportunities for research publication

Qualifikationen

• Proven experience training >1B parameter models.
• Strong Python skills with tests, CI/CD.
• Experience in open-source contributions valued.

Aufgaben

• Design scalable LLM pipelines and lead model development.
• Contribute to safety alignment and support data audits.
• Publish and present research at top venues.

Kenntnisse

Python engineering skills

Expertise in PyTorch

Clear scientific communication

Tools

DeepSpeed

HuggingFace Transformers

FSDP

Jobbeschreibung

Mission

We’re hiring a technically exceptional and impact-focused Postdoctoral Researcher in LLM engineering to lead research and development within the Meditron initiative—a suite of evolving open-source medical LLMs and multimodal foundation models.

You will contribute to model training, data curation, safety alignment, and global health deployments. You’ll join a fast-moving, interdisciplinary team of researchers, engineers, and students advancing trustworthy AI for medicine and humanitarian response.

Main Duties And Responsibilities

• Design scalable LLM pipelines using FSDP, DeepSpeed, HuggingFace Accelerate
• Lead model development (e.g., LLaMA, Mistral, Phi, Gemma) using LoRA, FlashAttention-2, MoE
• Contribute to safety alignment (RLHF, DPO, red-teaming, rejection sampling, calibration)
• Support data pipeline audits (tokenizer design, deduplication, privacy, synthetic supervision)
• Benchmark across general and medical tasks (lm-eval-harness, HELM, guideline adherence)
• Publish and present research at top venues (e.g., NeurIPS, ICLR, ML4H)
  
  

Collaborate closely with clinicians and humanitarian partners to ensure safety and usability

Profile

• Proven experience training >1B parameter models with distributed infrastructure
• Expertise in PyTorch, HuggingFace Transformers, DeepSpeed, FSDP
• Deep understanding of transformer internals and optimization strategies
• Strong Python engineering skills (tests, containers, CI/CD, reproducibility)
• Clear scientific communication and publication experience
  
  

Preferred

• Familiarity with clinical LLMs or decision support systems
• Experience with safety-critical evaluation (e.g., hallucination detection, benchmark leakage)
• Contributions to open-source projects
• Passion for equity-centered deployment and global health
  
  

Our Stack

PyTorch, HuggingFace, DeepSpeed, FSDP, WANDB, Hydra, MLFlow, WebDataset, Slurm, Docker, lm-eval-harness, OpenCompass, MedQA, SwissAlps (CSCS)

We offer

At LiGHT, we combine scientific rigor with purpose-driven action. We value creativity, humility, collaboration, and principled research that leads to tangible health impact. Our team culture embraces diverse backgrounds and sustained commitment to excellence, equity, and integrity.

Informations

Only applications submitted through the online platform are considered. You are asked to supply:

• A brief cover letter (pdf, up to 2 pages).
  
  

And In One PDF

• A CV with a publication list.
• A research statement (pdf, up to 3 pages).
• Contact details for 3 referees.
  
  

For any further information, please contact: [XXX]

Contract Start Date

Activity Rate : 100.00

Contract Type: CDD

Duration: 1 year, renewable

Reference: 1635