Project Scope & Activities
- Training Data Curation & Preparation:
- Designing and implementing strategies for collecting and annotating high-quality training data specific to industrial knowledge graph query generation.
- Working with domain experts to ensure the accuracy and relevance of the curated data.
- Developing scripts and tools in Python to automate data cleaning, transformation, and formatting for model training.
- Ensuring data privacy and compliance standards are met during curation.
- Model Fine-tuning:
- Experimenting with various fine-tuning techniques for pre-trained language models (LLMs) on the curated datasets.
- Utilizing cloud AI platforms (Google Cloud AI Platform, AWS SageMaker, Azure Machine Learning) for model training and deployment.
- Monitoring training progress, analyzing model performance metrics, and iterating on fine-tuning strategies.
Expected Outcomes
- Successfully curated a high-quality dataset suitable for fine-tuning LLMs for knowledge graph query generation.
- Contributed to the fine-tuning of LLMs on major cloud platforms (Google, AWS, Azure), leading to improved query generation capabilities.
- Provided actionable insights and recommendations to the product based on data analysis and model performance.
- Gained significant practical experience in applied machine learning, data engineering for AI, and cloud computing environments.
- Authored clear documentation on data curation processes, fine-tuning experiments, and evaluation methodologies.
Required Skills & Qualifications
- Machine Learning / Artificial Intelligence: Solid theoretical understanding of machine learning concepts, including natural language processing (NLP) and large language models.
- Python: Advanced proficiency in Python for data manipulation, scripting, and ML framework utilization (e.g., TensorFlow, PyTorch, Hugging Face Transformers).
- Data Analysis: Experience with data analysis libraries (e.g., Pandas, NumPy) and data visualization.
- Problem-Solving: Strong analytical and problem-solving abilities, with a methodical approach to data challenges.
- Collaboration: Ability to work effectively in a team, communicate technical concepts clearly, and adapt to evolving project requirements.
Bonus Skills (Nice to Have):
- Experience with cloud platforms (Google Cloud Platform, AWS, Azure) for ML workloads.
- Familiarity with knowledge graphs, graph databases, or semantic web technologies.
- Experience with MLOps practices or experiment tracking tools.
- Understanding of prompt engineering and agent design principles.
Top Skills
What We Do
Cognite is an AI company that delivers industrial software to improve the production efficiency of Energy, Process Manufacturing, and other industrial companies.
We deliver an Industrial DataOps platform that liberates siloed data and empowers our customers to solve some of their most complex business problems with AI-powered solutions. The typical solutions we enable drive innovative new ways to approach Data Exploration, Digital Operator Rounds, Production Optimization, Turnaround Planning, and Root Cause Analysis.
We do this by automating and scaling industrial data contextualization of various sources (such as time series, engineering diagrams, equipment logs, maintenance records, 3D facility models, images, large point clouds, and more). We use AI and other tools to find and map the meaningful relationships between the data across these various sources. In addition, we provide intuitive tools that enable efficient use of analytics and automated workflows, as well as prebuilt AI capabilities and a low-code industrial agent builder, Cognite Atlas AI, that enables AI to carry out more complex operations with greater accuracy.
Why Work With Us
Employees at Cognite are pushing the envelope with the latest cloud technology, scaling industrial applications across hundreds of assets, revolutionizing industrial data models, and working with robotics. Cogniters are fast, creative, and resilient. We keep the energy high and fun, learning from our mistakes and celebrating our victories together.
Gallery








