AI Engineer
GTMfund
Joining Arintra means working on a hard, real-world AI problem with a small, high-ownership team. You will build systems that make engineering faster, quality stronger, and production more reliable. We value structured thinkers, high-agency builders, and people who can convert messy workflows into simple, scalable systems. About the role:
You will build production-grade internal AI agents that reduce repetitive engineering work and improve reliability. This is a hands-on engineering role focused on building production AI systems that engineers rely on every day. The person must be able to design, ship, evaluate, monitor, and iterate agents used daily by engineering teams for PR review, ticket triage, test generation, go-live validation, customer call summarisation, regression checks, incident response, anomaly detection, and post-mortems.
Responsibilities
- Agent Development: Build and maintain internal agents for PR review, issue triage, testing, go-live validation, customer calls, regression checks, incident response, anomaly detection, and post-mortem drafting. Design each agent with clear success metrics, feedback loops, monitoring, cost tracking, and kill criteria. Create reliable agent workflows with structured outputs, tool use/function calling, retries, fallbacks, and audit trails. Measure adoption, quality, false positives, latency, and time saved for every agent.
- Agent Platform and Integrations: Build integrations with GitHub, Slack, Jira/Linear, Notion, PagerDuty, CI/CD systems, and BigQuery. Own webhook/event-driven triggers for PRs, issues, deploys, alerts, and transcript availability. Create reusable patterns for prompt management, prompt versioning, evaluation datasets, and model selection. Keep LLM usage cost-disciplined through caching, routing by model tier, context management, and per-agent cost visibility.
- SRE AI Safety Net: Build incident response agents that query logs, deploy events, error spikes, and accuracy signals to generate first-line triage within minutes. Build post-mortem and anomaly detection agents using BigQuery-backed reliability data. Design safe runbook automation where irreversible production actions require human approval. Partner with SDET and DevEx engineers to define the quality bar and data foundation for SRE-AI workflows.
- 3+ years of software engineering experience with at least 1 year building production LLM-powered systems or agents used by real users.
- Strong Python engineering skills; ability to write maintainable, tested, production-quality services.
- Hands-on experience with LLM APIs, preferably Anthropic Claude, including tool use/function calling, structured outputs, streaming, prompt caching, and multi-turn context handling.
- Experience with at least one agent/orchestration framework such as LangChain, LlamaIndex, DSPy, CrewAI, or strong custom orchestration experience.
- Strong REST API and webhook integration experience across tools such as GitHub, Slack, Jira/Linear, Notion, PagerDuty, or similar.
- Good understanding of LLM evaluation: defining success metrics, measuring false positives/negatives, creating feedback loops, and iterating from telemetry.
- Comfortable with Docker, CI/CD, observability, logging, alerts, and operating services in production.
- Strong SQL and BigQuery ability for querying logs, deploy events, incident timelines, accuracy signals, and time-series reliability data.
- Clear written communication; can write design docs that define scope, non-goals, evaluation methodology, owners, and risks.
- Experience with vector databases or retrieval systems such as pgvector, Pinecone, Weaviate, Chroma, or equivalent.
- Experience building engineering productivity, DevEx, or SRE tooling.
- Exposure to healthcare, medical coding, or high-accuracy regulated workflows.
- Experience with incident management systems, runbooks, or production on-call workflows.
- Experience running internal demos, enablement sessions, or playbooks for engineering teams.