- Company Name
- Lila Sciences
- Job Title
- Machine Learning Scientist, Open-Endedness (Level Flexible)
- Job Description
-
**Job Title**
Machine Learning Scientist, Open‑Endedness
**Role Summary**
Develop and advance generative AI systems that enable autonomous, continuous scientific discovery. Apply large‑model techniques (LLMs, diffusion, multimodal) and quality‑diversity (QD) algorithms to produce novel, high‑interestingness scientific hypotheses and designs, and devise unconventional evaluation and interpretability methods.
**Expectations**
- Lead research projects in open‑ended ML, from conceptualization to deployment.
- Publish in top AI/CS conferences (NeurIPS, ICML, ICLR, AAAI, GECCO, ICCC).
- Maintain high quality standards while exploring unconventional model pipelines.
- Collaborate cross‑functionally in a fast‑moving, unstructured environment.
**Key Responsibilities**
1. Design, implement, and iterate generative models (LLMs, diffusion, multimodal) for scientific exploration.
2. Develop and apply unconventional evaluation frameworks, including subjective assessments of interestingness.
3. Construct mechanistic interpretability tools and visualizations for large model internals.
4. Engineer and integrate QD techniques (MAP‑Elites, novelty search, POET, OMNI, minimal criterion novelty search, etc.) to generate diverse scientific propositions.
5. Train and supervise distributed ML workloads on cloud platforms (AWS, GCP, Azure) or on‑prem clusters.
6. Document methodologies, results, and best practices for internal and external dissemination.
**Required Skills**
- Strong foundation in deep learning frameworks (PyTorch, TensorFlow, or JAX).
- Proven ability to implement QD or neuroevolution algorithms.
- Experience with large‑scale distributed training and cloud infrastructure.
- Advanced research acumen in generative modeling, RLHF, distillation, or related areas.
- Excellent programming, debugging, and performance‑tuning skills in Python.
- Ability to communicate complex ideas clearly to multidisciplinary teams.
**Required Education & Certifications**
- PhD in quantitative discipline (Computer Science, Machine Learning, Physics, Chemistry, or related field) preferred.
- Self‑taught researchers with outstanding achievements and strong publication record will be considered.
- No mandatory certifications required.
---