Senior AI Platform/System Engineer, Coding Agents and MCP
At a glance
Senior ML Platform & Ops role at Pixalate. Remote, Pakistan · full-time.
Pay not stated ML Platform & Ops salary
remoteroles summary, based on the employer's posting.
What you'll do
- Turn product-specific coding agents into versioned cloud services, starting with MRT, and make them available to engineers
- Refresh agent knowledge from Bitbucket code and documentation after merges, then run evaluations for each codebase
- Build shared MCP services with per-user authentication, temporary credentials, limited permissions and audit trails
- Run AWS and GCP production systems, handling monitoring, incident fixes, patching, recovery, releases and infrastructure as code
- Control cloud spend and model usage, track costs by engineer and agent, and support adoption with onboarding and documentation
What you bring
- A computer science, software engineering or related degree, plus at least six years in DevOps, SRE or platform engineering
- Experience deploying shared AI coding agents for a team, including agents that run from CI or a server
- Hands-on skills with containers, infrastructure as code, configuration management, CI/CD, Linux, networking and production troubleshooting
- Experience securing AWS or GCP access with secrets management and least-privilege permissions, plus Python and shell scripting
- Cambridge O or A Levels, or an equivalent qualification, and strong written and spoken English
Who this fits
This senior role suits someone who has operated production systems and deployed shared coding agents, rather than only personal agent setups. You’ll work with engineering teams across the US, Singapore and Pakistan, with regular overlap during US business hours. Production on-call covers business hours, with an escalation contact for incidents outside that window. The role is based in Lahore and is remote.
From the employer
Senior AI Platform/System Engineer, Coding Agents and MCP
Employment Type: Full-Time
Location: Lahore, Pakistan (Remote)
About the Role
Our engineers write code with AI coding agents every day. Each product team has built its own Claude Code setup, loaded with that product's code, business rules and workflows. Today those agents live on individual laptops, so the next engineer on the same product starts from scratch.
We are hiring a Senior AI Platform/System Engineer in Lahore to turn them into shared agents that run in the cloud, and to run the cloud they run on. You will package each product's agent once, starting with our Media Ratings Terminal (MRT), host it for every engineer, keep its knowledge current as the code changes, and give people secure, measured access to it. Product leads own what each agent knows; you own the platform. You will also own our production infrastructure on AWS and GCP: monitoring, patching, backups and recovery, CI/CD, releases and cloud cost.
You will work with engineering leads across our product teams in the US, Singapore and Pakistan. We use Claude through managed APIs; the role does not include model training, data science or GPU serving.
Responsibilities
- Package each product's coding agent as a versioned plugin (skills, subagents, CLAUDE.md context and MCP configuration) and host it for every engineer on GCP and AWS with the Claude Agent SDK and headless Claude Code in containers (Cloud Run, ECS), triggered from Bitbucket Pipelines, Jira and Slack.
- Keep each agent accurate as the code changes: index code and documentation from our Bitbucket repositories and docs site, refresh it on every merge, and run evaluation suites per codebase, starting with MRT.
- Build and run shared MCP servers for Bitbucket, Jira, read-only BigQuery and AWS, with per-user authentication, short-lived per-session credentials, least-privilege access and audit logs, so no agent ever sees a raw key.
- Route model traffic through the Claude API, Amazon Bedrock or Google Vertex AI with budgets and cost caps, track usage and cost per engineer and per agent, and help engineers adopt the platform with documentation and onboarding.
- Operate our production cloud on AWS and GCP: configure and tune services, own monitoring and alerting (CloudWatch, Google Cloud Monitoring and our logging stack), and investigate and fix production issues with the engineering team.
- Take production on-call during business hours, with an escalation contact for incidents outside them.
- Keep systems patched and recoverable: OS and software upgrades, kernel and security patching, backups and disaster recovery, with restores tested.
- Own CI/CD on Bitbucket Pipelines and the release process across environments, and manage infrastructure as code and configuration management (Terraform, Ansible).
- Track and reduce cloud spend across AWS and GCP, including BigQuery, with budgets and regular reporting, and keep security hygiene current: IAM reviews, secrets management and access cleanup.
Qualifications
- Bachelor's degree in Computer Science, Software Engineering or a related field.
- 6+ years in DevOps, SRE or platform engineering, running production systems on AWS or GCP, including monitoring and alerting, patching, and backups and disaster recovery.
- Hands-on experience deploying AI coding agents (Claude Code or similar) for a team, not only for yourself: shared context such as CLAUDE.md or AGENTS.md files, skills, subagents or plugins, and agents run automatically from CI or a server. You will be asked to walk us through one.
- Containers, infrastructure as code and configuration management (Docker, Terraform, Ansible or similar), and CI/CD pipelines.
- Strong Linux administration, networking fundamentals and production troubleshooting.
- Secrets management and least-privilege IAM on AWS or GCP: secret managers, short-lived credentials and scoped service accounts.
- Strong scripting in Python, plus shell.
- Cambridge O or A Levels, or equivalent.
- Excellent written and verbal communication skills in English: you will write runbooks, design notes and updates for engineers in the US and Singapore. Assessed with a written exercise during the hiring process.
- Comfortable working across time zones, with regular overlap with US business hours.
Nice to Have
- A degree from an international university.
- Production experience with the Claude Agent SDK, headless Claude Code, or Claude Code in CI.
- Built or hosted MCP servers, including remote MCP servers with OAuth.
- Amazon Bedrock or Google Vertex AI as a model provider.
- Both AWS and GCP; Bitbucket Pipelines or GitLab CI.
- BigQuery administration and cost control.
- OpenTelemetry, LLM tracing or per-request cost tracking.
- Evaluation harnesses for agents or LLM output, such as golden datasets and regression runs in CI.
- Sandboxed execution for agents (containers, gVisor or Firecracker).
- TypeScript.
- Some exposure to running open-weight models (for example Ollama or vLLM). Not required.
About Us
Pixalate is an online trust and safety platform that protects businesses, consumers and children from deceptive, fraudulent and non-compliant mobile, CTV apps and websites. Our software and data have been used to unearth multiple high profile criminal and illegal surveillance cases including:- Gizmodo: An iCloud Feature Is Enabling a $65 Million Scam, New Research Says
- Adweek: A 7-Figure Ad Fraud Scheme Running on Roku Underlines Murkiness of CTV
- Washington Post: Your kids’ apps are spying on them
- Pro Publica: Porn, Piracy, Fraud: What Lurks Inside Google’s Black Box Ad Empire
- ABC7 News: The State of Children's Privacy Online
- NBC News: How many apps are tracking your children