10 hours ago
Campinas, BrazilSenior
Responsibilities
- Design, develop, and deploy multi-agent AI systems for reasoning, planning, and task execution.
- Implement RAG pipelines using LangChain, LangGraph, and Azure AI Foundry.
- Optimize model serving for low-latency and high-throughput inference using KServe, Triton Inference Server, or Ray Serve.
- Build observability and evaluation frameworks to monitor agent reasoning, success rates, and failure cases.
- Integrate structured and unstructured data sources into agent workflows with ML and data engineers.
- Apply AI security and alignment techniques, including guardrails, prompt-injection prevention, and red-teaming.
- Deploy and iterate on systems in a CI/CD environment using Azure DevOps, GitHub Actions, and ArgoCD.
- Contribute to architecture decisions involving distributed systems, GPU usage, caching strategies, and memory management.
Requirements
- Strong proficiency in Python and C++ or Go for performance-critical components.
- Proven experience with LLMs such as OpenAI, Anthropic, Llama, or Mistral.
- Practical experience with LangChain, LangGraph, or Semantic Kernel.
- Deep understanding of search and retrieval systems and vector databases such as Pinecone, Weaviate, FAISS, or Milvus.
- Knowledge of distributed systems and experience deploying ML models in Kubernetes, AKS, EKS, or GKE environments.
- Familiarity with ONNX Runtime, TensorRT, or DeepSpeed for inference optimization.
- Experience with Azure, AWS, or GCP and infrastructure-as-code tools such as Terraform, CloudFormation, or Bicep.
- Understanding of AI security, including prompt injection, data leakage, and adversarial testing.
- Experience with multi-agent simulations such as AutoGen, CrewAI, or OpenAI Swarm is preferred.
- Background in AI-model red-teaming and offensive security is preferred.
- Knowledge of graph databases such as Neo4j or ArangoDB is preferred.
- Publications, research contributions, or open-source involvement related to LLMs or agent frameworks is preferred.
Benefits
- Performance-based bonus, attendance bonus, private pension plan, meal allowance, health, dental, and life insurance.
- Days off, medicines discounts, WellHub partnership, childcare subsidies, and discounts on Ambev products.
- Clube Ben partnership, scholarship, school materials assurance, language and training platforms, and transport allowance.
- Casual office and dress code; applicable rules apply to marked benefits.
Tech Stack
Categories
About BEES
BEES builds a B2B digital commerce and marketplace platform for small and medium retailers to order products, manage invoices, and track deliveries from AB InBev and partner suppliers. The business supports online ordering, promotions, and data-driven sales tools across 29 countries. Created by and owned by AB InBev, BEES is headquartered in Rotkreuz, Zug, and focuses on digitizing distributor-to-retailer workflows in beverage and adjacent categories.