Employment Type: 6 Month Contract

We are seeking a highly skilled LLM Training Engineer to join our team, with a focus on developing and refining content for post-training phases of large language models (LLMs). The ideal candidate will play a crucial role in curating, generating, and managing high-quality datasets, prompts, and scenarios used for fine-tuning and optimizing LLM performance after the initial pre-training stages. You will collaborate closely with machine learning engineers, data scientists, and domain experts to ensure that LLMs continue to improve in their specific tasks and domains.

Key Responsibilities:

– Content Creation and Curation: Develop, refine, and curate high-quality content (code, text, prompts, dialogues, scenarios) for use in the post-training of large language models optimized for code generation.
– Fine-tuning Data: Design datasets for tasks such as instruction following, question answering, reasoning, conversational models, etc., to improve model accuracy, robustness, and generalization.
– Post-training Optimization: Collaborate with ML engineers to integrate the curated content into fine-tuning pipelines, ensuring that models are tailored to specific use cases.
– Evaluation and Benchmarking: Create scenarios to evaluate LLM performance and establish benchmarks for targeted improvements in different areas such as comprehension, creativity, factuality, and bias reduction.
– Feedback Loop Development: Set up human-in-the-loop feedback systems for continuous model improvement, and work closely with annotators or SMEs to enhance dataset quality.
– Data Quality Control: Monitor data quality, diversity, and relevance; identify and resolve gaps or biases in the training data.
– Collaboration: Work with cross-functional teams including product managers, data scientists, and engineers to align post-training content with business goals and technical requirements.
– Documentation: Write and maintain clear and detailed documentation on content curation methodologies, dataset specifications, and post-training processes.

Required Qualifications:

– Educational Background: Bachelor’s or Master’s degree in Computer Science, Data Science, Computational Linguistics, or a related field.
– Experience: 3+ years of experience working with large language models or machine learning frameworks, with a focus on post-training or fine-tuning.
– Programming Skills: Proficiency in Python and experience with ML frameworks such as PyTorch or TensorFlow.
– Data Handling: Experience in curating, cleaning, and managing large datasets for machine learning applications.
– Collaborative Mindset: Strong communication skills and the ability to work effectively in a cross-functional team.
– Problem Solving: Strong analytical and problem-solving abilities with a focus on improving model performance through targeted post-training interventions.

Preferred Qualifications:

– Experience working with LLM architectures such as GPT, BERT, T5, or similar.
– Knowledge of reinforcement learning from human feedback (RLHF).
– Knowledge of fine-tuning techniques such as Low-Rank Adaptation.
– Familiarity with tools for annotation, labeling, and dataset management.
– Understanding of bias and fairness issues in AI, and experience in mitigating these through training.

About DFINITY and the Internet Computer:

DFINITY is a leading contributor to the Internet Computer Protocol (ICP), with a mission to bring the world’s compute onto the secure ICP network. Built on its unique third-generation blockchain technology, ICP enables the development and operation of a new generation of unstoppable, tamper-proof, fully decentralized web applications. Its powerful technology can run entire AI models within smart contracts, representing a major advancement for secure AI. Through seamless integration with Bitcoin, Ethereum, and other networks, ICP facilitates multi-chain operations for digital assets and web3.

Join our team of over 250 talented individuals, including world-renowned cryptographers, distributed systems engineers, programming language experts, and industry leaders, who are shaping the future of the internet and web3.

DFINITY was founded in 2016 by entrepreneur and crypto theoretician, Dominic Williams.

All qualified applicants will receive consideration for employment without regard to race, color, religion, gender, gender identity or expression, sexual orientation, national origin, genetics, disability, age, or veteran status.