AI Engineer

Overview

Own an AI agent end to end within Certa's third-party risk management platform, owning its harness (agent loop, tools, context, evals, model routing and backend services) and measured on task success, latency, cost per task and customer outcomes.

What you'll do

• Own the agent loop: step control, planning, retries, error feedback, subagents, and parallel tool calls for long, multi-step tasks.
• Design tools and actions: build the tools, MCP servers, and skills agents use, narrow, well-described, permissioned, and safe to call against customer data.
• Engineer context: decide what the model sees on every step, including compaction, memory, retrieval-as-a-tool, and how agents read large enterprise records.
• Build evals that gate every change: golden sets, regression suites, and trace-based scoring wired into CI.
• Own model choices and the runtime: route across models on capability, latency, and cost, with durable sessions, sandboxed execution, nested-trace observability, token budgets, and graceful fallbacks for multi-tenant enterprise customers.
• Own the backend systems: design services end to end (data model, APIs, storage choice, failure modes), build async pipelines (queues, workers, retries, idempotency, backpressure), and make them fast and cheap at scale on multi-tenant AWS with isolation, least-privilege access, PII handling, and audit trails.
• Own it in production: write the design doc, weigh trade-offs, defend them in review, and own the service in production.

What you'll need

• 3+ years building production backend systems with strong system design, including designing and scaling services under real load, picking data stores, modeling data, and handling async processing, queues, caching, idempotency, and failure in production.
• Shipped at least one LLM agent to real users and can explain its loop, tools, failure modes, and the numbers that showed it worked.
• Built evals that changed a decision: killed, shipped, or reworked something because the eval said so.
• Strong instincts for context and tool design, able to say why an agent failed by reading its trace.
• Uses coding agents (Claude Code, Codex, Cursor) as a core part of how they work, with opinions about what makes a harness good.
• A record of owning outcomes: picks the metric, moves it, and reports honestly when it doesn't move.

Nice to have

• Built or contributed to an agent harness, framework, or orchestration layer (open source counts).
• Enterprise B2B experience in a regulated domain (PII, audit trails, permissions, SOC 2).
• Founder or founding-engineer experience.
• Experience with any framework (LangGraph, Agent SDKs, Deep Agents) or none.

Details

• Python preferred.
• Remote-first, hiring globally; optional office locations in London, San Francisco and Jaipur.
• Compensation: top of market salary plus bonuses and equity; 49 days paid time off; full benefits, retirement and wellness plans; professional development support and remote workspace setup bonus; annual company offsites.