Artificial Intelligence Engineer

BULL-IT SOLUTIONS LTD; México; On-site; MXN 900,000 - 1,300,000; Full time

Job summary

BULL-IT SOLUTIONS LTD is seeking an experienced ML/LLM engineer to design, build and support production ML/LLM systems used by real users in Mexico. You will work on GenAI engineering, agentic automation, and robust evaluation processes, with emphasis on reliable, observable deployments and incident response.

The role requires hands-on Python production code, experience with ML libraries, and strong RCA/AIOps capabilities.

Qualifications

• 3+ years designing production ML/LLM systems for real users.
• Experience with GenAI/LLM engineering including RAG, embeddings, vector search, and tool calling.
• Experience with agentic AI, LangGraph/LangChain, guardrails, and auditing.
• Defined evaluation sets and metrics; track improvements based on results.
• AIOps or anomaly detection across metrics/logs/events.
• Proficient Python with tests; common ML libraries listed above.
• Observability tools such as Splunk plus Prometheus/OpenTelemetry/Dynatrace.
• Root-cause analysis and RCA tooling experience.

Skills

3+ years ML/LLM production systems

GenAI/LLM engineering topics

Agentic AI and automation

Evaluation and metrics

AIOps and anomaly detection

Production Python with tests

Observability and monitoring

Root-cause analysis tooling

Tools

Pandas

NumPy

scikit-learn

PyTorch

TensorFlow

Job description

• 3+ years designing, building, and supporting production ML or LLM systems used by real users; POCs, demos, and coursework alone are not sufficient.
• GenAI/LLM engineering: RAG, chunking, hybrid search, reranking, embeddings, vector search, tool/function calling, structured outputs, and handling incorrect or unsupported answers.
• Agentic AI: agents or multi-step automation using LangGraph, LangChain, Semantic Kernel, or similar, including tools, guardrails, human approval, and action auditing.
• Evaluation: defined evaluation sets and metrics, measured accuracy or error rates, and system improvements based on results.
• AIOps/anomaly detection: time-series anomaly detection and alert-noise reduction across metrics, logs, or events.
• Python production code with tests; relevant libraries may include Pandas, NumPy, scikit-learn, PyTorch, or TensorFlow.
• Production troubleshooting during live incidents across Linux (CPU, memory, disk, network), Java (heap, garbage collection, thread dumps), and Python.
• Observability: Splunk (SPL, dashboards, alerts) plus at least one of Prometheus, OpenTelemetry, Dynatrace, or a similar tool.
• RCA/AIOps tooling: significant contribution to root-cause analysis, AIOps, or incident-analysis tooling.