Senior AI Engineer with AZURE AI experience to solution design and work on an Innovative project harnessing AI and Data analytics - 0128658

Our Public enterprise client is looking for a Senior AI Engineer with AZURE AI experience to solution design and work on an Innovative project harnessing AI and Data analytics - 0128658

1-year contract, possibilities of extension; Vancouver based, Hybrid work model with the requirement to be in-office a minimum of twice per month and more if needed.

Must Have:

• Software Development/AI Engineer experience on cloud (Azure) within DevOps and Agile environment
• Proficiency in programming languages like Python, which is widely used for AI and machine learning, or other languages commonly used in AI.
• Graph Theory & Knowledge Graphs
• Experience building and querying knowledge graphs to represent structured knowledge and relationships, particularly in the context of enhancing retrieval-augmented generation (RAG) models.
• Familiarity with graph databases such as Neo4j, ArangoDB, or Amazon Neptune is valuable for structuring and storing graph data.
• Retrieval-Augmented Generation (RAG)
• Expertise in RAG models, which combine retrieval-based and generative approaches to improve AI's ability to generate contextually relevant and accurate responses by retrieving data from external sources like documents, databases, or web data.
• Experience implementing or fine-tuning models such as RAG from Facebook AI, which integrates retrieval-based techniques (e.g., Dense Retriever, BM25) with generative models (e.g., GPT, T5).
• Information Retrieval and Search Algorithms
• Deep knowledge of information retrieval (IR) techniques for implementing document or knowledge retrieval systems that form the basis of RAG.
• Experience with embedding-based search, nearest neighbour search, and indexing techniques using libraries like FAISS, Annoy, or Elasticsearch to retrieve relevant information from large datasets.
• Graph Data Processing & Manipulation
• Ability to preprocess and manipulate graph data for tasks like entity resolution, graph pruning, and graph-to-text generation.
• Knowledge of graph-based algorithms such as PageRank, centrality measures, and community detection to enhance the accuracy of generative AI models based on graph data.
• Large-Scale Model Training & Optimization
• Experience with distributed training techniques for large generative models, especially when using graph data that may require massive computational resources.
• Familiarity with optimization methods, including gradient-based optimization and multi-task learning, to fine-tune AI models that involve both graph learning and language generation.
• Experience with techniques for optimizing model performance for deployment, including hardware acceleration (e.g., GPUs/TPUs), pruning, and quantization.
• Embedding Models & Vector Representations
• Expertise in building and fine-tuning graph embeddings and sentence embeddings that capture semantic relationships within graph-based structures, improving the quality of downstream generative AI tasks.
• Proficiency with vector databases and embedding management tools like CosmosDB or Pinecone for efficiently querying and using vector-based representations of graph data.
• Model Evaluation & Metrics
• Strong understanding of performance metrics for generative AI models, including BLEU, ROUGE, perplexity, and generation quality.
• Ability to evaluate retrieval-augmented models through both precision/recall of retrieved information and relevance of generated output, ensuring that the model produces contextually relevant results when using graph data.
• Diversity Metrics: In generation tasks, metrics such as distinctness or novelty measure how diverse the generated outputs are, helping assess diversity and creativity of generation.
• Knowledge of deployment strategies such as containerization (AKS), microservices, and serverless architectures for deploying generative AI applications that leverage graph data for real-time use cases.
• ETL (Extract, Transform, Load): Knowledge of ETL processes for data wrangling, preparing data for graph databases, and integrating various data sources
• Undergraduate degree in Computer Science or STEM (Science, Technology, Engineering, Mathematics). An equivalent combination of education/experience

Nice to Have:

• Experience with Data Science
• Exposure to / experience with Advanced RAG techniques such as Lazy GraphRAG, DRAG (Dynamic RAG), MultiHop RAG for increasing accuracy and reducing costs.