Acceleration Center - Agentic AI & DevOps Engineer - Senior Associate

PwC; Monterrey; Presencial; MXN 600.000 - 900.000; Jornada completa

Descripción de la vacante

PwC Acceleration Center Mexico seeks an Acceleration Center Agentic AI Engineer — Senior Associate to transform raw data into actionable insights and drive business growth across client portfolios, leveraging Agentic AI, RAG, and LLMs. You will oversee model layer end-to-end, from prompt design to MLOps, ensuring scalable, cost-efficient production.

You will work with global teams across Advisory, Assurance, Tax and Business Services, building observability, evaluation frameworks, and CI/CD for

Formación

• At least a Bachelor’s degree.
• At least 2 years of experience.
• Oral and written proficiency in English required.

Responsabilidades

• Design and deploy LLM-powered automations at scale - agents, skills, pipelines - from prototype through production.
• Build eval frameworks: offline benchmarks, adversarial red‑team­ing, regression suites, and production feedback loops.
• Own observability end-to-end: distributed traces, latency histograms, cost attribution, token throughput, error budgets, and dashboards that surface signal over noise.
• Define and enforce SLOs for model-backed services; own the alerting stack and incident response playbooks.
• Debug production failures across the full stack - from Kubernetes pod restarts to model hallucinations to prompt injection - and write postmortems that actually prevent recurrence.
• Operate AI-natively: use Claude Code, Codex, GitHub Copilot, Cursor, and similar tools as daily infrastructure, not novelty - and hold the output to the same standard as hand‑written code.
• Instrument and monitor LLM behavior in production: token usage, response quality, failure modes, latency SLAs Architect CI/CD for model and prompt changes: versioning, automated eval gates, rollback strategies.
• MLOps/SRE: shipped LLM‑backed automations to productions with real SLAs and real users, reproducible experiments (tracked, versioned, auditable), SLO definition, error tracking, alerting design, monitoring pipelines for model degradation, prompt drift and cost anomalies.
• AI Native Fluency: daily user of AI coding assistants (Claude Code, Codex, Copilot, Cursor), understand in‑model limitations, failure modes and when to distrust output, can write prompts and agent instructions with rigor, has opinions on eval design, context management and tool‑use patterns.

Conocimientos

Python

FastAPI

Async patterns

Pydantic

Typescript

SQL/Redis

Kubernetes

Azure DevOps CI/CD

Claude SDK

OpenAI SDK

Langchain

Langgraph

Langfuse

GitHub Copilot

Cursor

Educación

Bachelor's degree

Herramientas

Claude Code

Codex

GitHub Copilot

Cursor

Langchain

Langgraph

Langfuse

Descripción del empleo

Industry/Sector

Not Applicable

Specialism

Data, Analytics & AI

Management Level

Senior Associate

Job Description & Summary

The Opportunity

Join our Acceleration Center Mexico and help shape the future of business for our diverse client portfolio across geographies and jurisdictions. You’ll work at the heart of global teams across Advisory, Assurance, Tax and Business Services- solving real client challenges through connected collaboration. We’ll help you grow your skills so you can go further. With hands-on learning, cutting-edge tools and an inclusive culture, this is your opportunity to do inspiring work that makes a difference—every day.

As an Acceleration Center - Agentic AI Engineer - Senior Associate, you will play a pivotal role in transforming raw data into actionable insights, enabling informed decision-making and driving business growth. Within our Internal Firm Services practice, you will leverage Artificial Intelligence to solve business problems through Agentic Systems, RAG, LLMs, while ensuring solution scalability and correct implementation.

As a Senior Associate, you will focus on building meaningful client connections and learning how to manage and inspire others. You will navigate increasingly complex situations, growing your personal brand and deepening your technical skills. You are expected to anticipate the needs of your teams and clients, delivering quality work while embracing increased ambiguity. This role requires you to respond effectively to diverse perspectives, use a broad range of tools and methodologies to generate new ideas, and uphold professional and technical standards.

In this role at PwC Acceleration Center Mexico, owns the model layer end-to-end: prompt and skill design, evals, safety, and the MLOps that keeps it cheap, fast, traced, and reproducible in production.

Responsibilities

• Design and deploy LLM-powered automations at scale - agents, skills, pipelines - from prototype through production

-Build eval frameworks: offline benchmarks, adversarial red‑team­ing, regression suites, and production feedback loops that catch drift before users do

-Own observability end-to-end: distributed traces, latency histograms, cost attribution, token throughput, error budgets, and dashboards that surface signal over noise

-Define and enforce SLOs for model-backed services; own the alerting stack and incident response playbooks

• Debug production failures across the full stack - from Kubernetes pod restarts to model hallucinations to prompt injection - and write postmortems that actually prevent recurrence

-Operate AI-natively: use Claude Code, Codex, GitHub Copilot, Cursor, and similar tools as daily infrastructure, not novelty - and hold the output to the same standard as hand‑written code

-Instrument and monitor LLM behavior in production: token usage, response quality, failure modes, latency SLAs Architect CI/CD for model and prompt changes: versioning, automated eval gates, rollback strategies

What You Must Have

• At least a Bachelor’s degree
• At least 2 years of experience
• Oral and written proficiency in English required

What Sets You Apart

• Core engineering: Python fluency, FastAPI, async patterns, Pydantic, Typescript, SQL/Redis for state and caching, K8s + Azure DevOps CI/CD
• ML/LLM Depth: strong fundamentals (transformers, finetuning, RAG, agent frameworks); hands‑on with Claude/OpenAI SDKs/Langchain/Langgraph/Langfuse; building evaluations: prompt tests, LLM‑as‑a‑judge pipelines, human feedback integration.

-MLOps/SRE: shipped LLM‑backed automations to productions with real SLAs and real users, reproducible experiments (tracked, versioned, auditable), SLO definition, error tracking, alerting design, monitoring pipelines for model degradation, prompt drift and cost anomalies.

-AI Native Fluency: daily user of AI coding assistants (Claude Code, Codex, Copilot, Cursor), understand in‑model limitations, failure modes and when to distrust output, can write prompts and agent instructions with rigor, has opinions on eval design, context management and tool‑use patterns.

Travel Requirements

Up to 20%

Job Posting End Date

octubre 23, 2026