Ingeniero de Base de Conocimiento e Inteligencia Artificial

This is a fully remote position, open to applicants in Colombia.

📋 Description

• Develop the strategy for knowledge and generative artificial intelligence for agents, encompassing RAG architecture, document preparation and segmentation, model selection and routing, quality assessment, performance metrics, cost optimization, and observability mechanisms.

• Transform official knowledge sources and use case requirements into a viable technical architecture for agents and conversational experiences.

• Create the conceptual architecture of the knowledge base for agents.

• Define the RAG strategy: document ingestion, extraction, cleaning, normalization, chunking, metadata, embeddings, indexing, semantic and hybrid retrieval, re-ranking, citation, and traceability.

• Establish quality requirements and the structure of official content.

• Design versioning, publication, depublication, and document updating processes.

• Differentiate between current, historical, contradictory, or pending approval content.

• Design the interaction between the knowledge base, moderator, agents, and language models.

• Propose selection criteria for models based on complexity, cost, latency, accuracy, and criticality.

• Create routing rules between models and agents.

• Optimize token consumption, latency, cost per conversation, response reuse, caching, and context.

• Define evaluation metrics for responses and retrieval, including accuracy, relevance, groundedness, faithfulness, coverage, abstention, hallucinations, latency, and cost.

• Design cases and scenarios for technical and functional evaluation.

• Establish guardrails for responses, tools, sensitive data, and agent actions.

• Develop observability mechanisms for prompts, models, tokens, responses, sources, and errors.

• Document risks associated with generative AI and model constraints.

• Collaborate with Data Architecture to define dependencies and the quality of sources.

• Work with Integration Architecture to define capabilities and tools consumable by agents.

• Participate in the conceptual design of the moderator and agent orchestration.

• Generate inputs for the technical backlog and subsequent implementation.

⛳️ Requirements

• Professional degree in Systems Engineering, Software Engineering, Computer Science, Artificial Intelligence, or related fields.

• Professional experience in data engineering, AI, machine learning, semantic search, or intelligent platform development.

• Hands-on experience with LLMs and generative AI applications.

• Practical experience with RAG, embeddings, vector databases, semantic and hybrid search, prompt engineering, model and response evaluation, and document processing.

• Knowledge of Python and AI model APIs.

• Experience with vector databases or search engines.

• Understanding of chunking, metadata filtering, re-ranking, and retrieval.

• Ability to design technical and functional quality metrics.

• Familiarity with observability, traceability, and cost optimization in AI solutions.

• Capability to document architectures, experiments, decisions, and evaluation criteria.

• Desirable: experience with Azure OpenAI, Azure AI Search, Databricks, Microsoft Fabric, Amazon Bedrock, Vertex AI, OpenSearch, Elasticsearch, Pinecone, Weaviate, or other equivalent technologies.

• Desirable: experience with LangChain, LlamaIndex, Semantic Kernel, or other frameworks.

• Desirable: knowledge of agents, tool calling, function calling, and multi-agent orchestration.

• Desirable: experience in OCR and intelligent document processing.

• Desirable: understanding of security for LLMs and OWASP Top 10 for LLM Applications.

• Desirable: experience in designing virtual assistants or conversational channels.

🏝️ Benefits

• Comprehensive health and wellness benefits.

• Opportunities for professional development and continuous learning.

• Flexible working hours and remote work options.

• Collaborative and innovative work environment.

• Competitive salary and performance-based bonuses.