## Role Overview We’re seeking an **AI Engineer** who will design and deploy intelligent systems that leverage large language models (LLMs), retrieval-augmented generation (RAG), and vector databases to solve complex enterprise problems. You will build end-to-end pipelines for ingesting, encoding, and indexing data; integrate knowledge graphs and hybrid retrieval strategies; and optimize models for latency, accuracy, and cost. --- ## Key Responsibilities ### Design, Prototype, and Deploy RAG Systems Architect scalable RAG pipelines that combine vector search, hybrid retrieval, re-ranking, and contextual compression techniques. Build and integrate vector search systems (e.g., Milvus, pgvector, FAISS, Weaviate) for high-recall retrieval across structured and unstructured data. ### Develop Hybrid Retrieval and Knowledge-Driven Pipelines Design hybrid retrieval systems that blend semantic, symbolic, and graph-based methods. Create custom chunking and encoding strategies to store operational knowledge in vector databases and knowledge graphs. ### Build and Integrate Knowledge Graphs Architect knowledge graphs (Neo4j, RDF, custom schemas) and integrate them into retrieval workflows to support reasoning and decision-making. ### Optimize Data Pipelines and Embeddings Build and optimize data pipelines that convert incoming documents into high-quality embeddings for AI retrieval. Tune chunk sizes, indexing frequencies, and embedding strategies to enhance recall, factual accuracy, and efficiency. ### Implement Hybrid Search and Metadata Filtering Combine semantic and keyword search to improve precision and efficiency. Experiment with metadata filtering techniques to surface the most relevant context for AI reasoning agents. ### Evaluate and Monitor System Performance Evaluate end-to-end retrieval performance using classical IR metrics (precision, recall) and LLM-specific evaluations (factuality, coherence, task success). Monitor retrieval logs and adjust embedding configurations to maintain relevance and mitigate hallucinations. ### Compare and Fine-Tune LLMs Compare performance of different LLMs (e.g., GPT-4, Claude, Llama) across embedding structures and refine tuning strategies. Implement quantization, distillation, and optimization techniques to meet latency, throughput, and cost targets. ### Collaborate and Enable Teams Work cross-functionally with product managers, data engineers, and domain experts to translate product goals into scalable AI solutions. Conduct workshops and enablement sessions to enhance AI literacy across internal teams. ### Ensure Quality and Compliance Participate in rigorous code reviews and implement testing frameworks to ensure reliability, security, and compliance. Continuously monitor model accuracy and safety, and uphold data governance and ethical guidelines. --- ## Required Qualifications ### Experience 3+ years of software engineering experience with deep expertise in Python. Proven experience building and deploying RAG or information-retrieval systems, with strong proficiency in TensorFlow and PyTorch. ### Retrieval Expertise Demonstrated ability to design hybrid retrieval pipelines, encode knowledge using LLMs and vector stores, and build and optimize RAG systems. ### Vector Databases and Search Algorithms Proficiency with vector databases and search libraries such as pgvector, FAISS, Milvus, Pinecone, or Weaviate. Strong understanding of vector search algorithms, indexing strategies, and hybrid search techniques. ### Embedding and LLM Frameworks Hands-on experience with embeddings and transformer-based models (e.g., OpenAI, Cohere, Sentence Transformers) and frameworks such as Hugging Face Transformers, LangChain, and LlamaIndex. ### Distributed Systems and Deployment Practical knowledge of distributed systems, ETL pipelines, Docker, and Kubernetes, along with cloud platforms (Azure, AWS, GCP) for deploying AI applications. ### Evaluation and Security Familiarity with evaluation of retrieval systems, observability tools, and model performance monitoring. Understanding of data governance, security, and compliance considerations. --- ## Preferred Qualifications ### Knowledge Graphs and Multimodal Search Experience designing and deploying knowledge graphs, semantic graphs, or multimodal search systems. ### Fine-Tuning and RLHF Familiarity with LLM fine-tuning, reinforcement learning from human feedback (RLHF), and safety alignment. ### Multimodal AI Exposure to multimodal models (image, video, audio) and diffusion models. ### Open-Source Contributions Contributions to open-source generative AI, retrieval, or vector database projects, or published research and technical blogs. ### Front-End Prototyping Experience with React or Next.js for rapid prototyping of AI-driven applications. ### Advanced Degrees Master’s or PhD in Computer Science, AI, Machine Learning, or related fields (preferred but not mandatory). Extensive relevant experience or significant open-source contributions may substitute formal education.