AI LLM Scientist

AI LLM Scientist

Company details

Description of the offer

We’re looking for an AI LLM Scientist who can push the boundaries of model intelligence, reasoning, and agentic capabilities. If you enjoy diving deep into transformer architectures, experimenting with frontier models, optimizing training pipelines, and prototyping novel ways LLMs can plan, act, and collaborate—this role will feel like home.

You’ll play a key part in shaping our next generation of AI-driven products: smarter agents, efficient pipelines, adaptive workflows, and robust AI behaviors built on top of cutting-edge research.

What You’ll Work On

1. Core LLM Research & Development
2. Experiment with foundational models (e.g., GPT, Claude, Llama, Mixtral, Phi).
3. Fine-tune, supervise, distill, and reinforce models for specialized tasks.
4. Improve reasoning, planning, and tool-use capabilities of agentic workflows.
5. Build evaluation frameworks for multi-step logic, safety, and reliability.
6. Research and deploy optimization techniques: LoRA, QLoRA, 4-bit quantization, SFT, RLHF/RLAIF, preference modeling, chain-of-thought optimization.
7. Agentic AI & System Design
8. Architect scalable pipelines for multi-agent or tool-augmented systems.
9. Prototype autonomous behaviors, memory systems, retrieval workflows, and reasoning chains.
10. Integrate external tools/APIs, knowledge bases, and embeddings for enhanced context.
11. LLM Engineering
12. Deploy models efficiently using GPU/TPU stacks.
13. Optimize latency, throughput, and cost for production workloads.
14. Experiment with model compression, routing, and hybrid LLM ensembles.
15. Research + Product Collaboration
16. Translate complex AI research into practical, reliable features.
17. Work closely with engineers, product teams, and designers to ship experimental ideas.
18. Stay ahead of the AI frontier and propose new directions for the team.

What You Should Bring

Strong understanding of LLMs, transformers, embeddings, and NLP fundamentals.
Hands-on experience with training/fine-tuning frameworks: PyTorch, JAX, Hugging Face, DeepSpeed, Ray.
Experience building or optimizing LLM-powered applications and agents.
Ability to read, interpret, and apply research papers (NeurIPS, ICML, ICLR, ACL, etc.).
Solid grasp of distributed systems, model optimization, or inference engineering.

Bonus points for:

Publications or open-source research contributions
Experience training smaller foundation models
Knowledge of RLHF pipelines
Exposure to retrieval systems, vector DBs, or tool-use frameworks
Familiarity with frontier safety research

Location of employment

How to apply?