This article outlines a hands-on engineering role focused on fine-tuning and evaluating open-source foundation models (LLMs and multimodal), building end-to-end model training pipelines including data preparation, labeling, and evaluation, optimizing models for latency and accuracy on GPU/cloud infrastructure, and implementing prompt engineering, model alignment, and personalization techniques. It also summarizes required experience, bonus skills, and what you'll gain.
Core responsibilities and technical focus
This role centers on practical model development and deployment: fine-tuning and evaluating open-source foundation models (LLMs & multimodal), constructing robust model training pipelines, and optimizing models to balance latency and accuracy on GPU/cloud infrastructure. You will implement prompt engineering, model alignment, and personalization techniques as part of shipping high-quality model behavior.
- Fine-tune and evaluate open-source foundation models (LLMs & multimodal)Work directly on model adaptation and assessment to achieve target behavior and performance.
- Build model training pipelinesDevelop end-to-end pipelines covering data preparation, labeling, and evaluation to support repeatable training and validation.
- Optimize for latency and accuracy on GPU/cloud infrastructureTune models and execution paths to meet latency and accuracy requirements when running on GPU or cloud resources.
- Implement prompt engineering, model alignment, and personalizationApply prompt design, alignment strategies, and personalization approaches to shape model outputs for real users.
Experience, bonuses, and what you'll gain
This section describes the experience you should bring, valuable bonus skills, and the outcomes you can expect from the role, including ownership and rewards.
- What you’ll bring
- Experience with open-source model ecosystems (Hugging Face, Llama, Mistral, etc.).
- Strong Python skills and hands-on PyTorch or TensorFlow experience.
- A demonstrated GitHub portfolio.
- Familiarity with APIs, vector DBs, RAG pipelines, and agent frameworks.
- Cloud (AWS/GCP/Azure) and GPU pipeline experience is a plus.
- Ability to rapidly prototype.
- Bonus
- Dataset capture/cleaning and multimodal architectures.
- UI frameworks such as Next.js/React.
- Hackathon, Kaggle, or open-source contributions.
- What you’ll gain
- Build and launch an LLM product used by customers.
- Mentorship from founders and engineers.
- Ownership of core features and research direction.
- Potential pre-placement offer, stipend, and performance incentives.
In summary, this role centers on fine-tuning and evaluating open-source foundation models (LLMs & multimodal), building model training pipelines (data preparation, labeling, evaluation), optimizing for latency and accuracy on GPU/cloud infrastructure, and implementing prompt engineering, model alignment, and personalization. Candidates with the listed experience can build and launch an LLM product, receive mentorship, ownership, and potential offers, stipend + incentives.
