Join beqom — where tech meets impact
beqom is a high-growth B2B SaaS company that provides industry-leading tools for pay equity and transparency, compensation, and performance management.
Trusted by some of the world’s most respected companies, beqom enables HR and business leaders to navigate global compliance and make smarter pay decisions that attract, retain, and motivate top talent.
Founded in Switzerland and serving clients worldwide, our powerful, enterprise-ready products are fuelled by beqom pay intelligence.
Role Overview
We are looking for a Senior Agentic Backend Engineer to join our Data Platform Core team, which owns the services at the heart of beqom’s data architecture — multi-tenant data ingestion, schema management, snapshots, calculated fields, and the data contracts powering analytics and reporting across the product.
In this role you will not write the code. AI agents will. You will own the specifications that govern these services, the agentic pipeline that turns those specifications into production software, and the verification machinery that proves the output is correct — with the explicit goal of reaching fully AI-embedded development: no human-written code, no human line-by-line review, and external behavioural scenarios, not traditional test suites, as the quality gate.
You will collaborate closely with Data Platform and Analytics engineers, Support teams, and Product stakeholders — translating their needs into agent-executable specifications. Specs are our system of record; agents do the typing.
What you'll be doing
Agentic Delivery
- Deliver features across multi-tenant data platform services — data ingestion, schema management, effective dating, snapshots, calculated fields, and exports — exclusively through AI agents working from your specifications.
- Author structured, testable Product Requirement Prompts (PRPs) and spec files as the primary engineering artefact, and maintain the versioned spec library that governs the codebase.
- Run multiple agents in parallel across workstreams; evaluate outcomes against specs; progress agent first-pass success rate, defect escape rate, and the share of changes merged without human review steadily upward.
Pipeline Engineering
- Build and maintain the agentic pipeline: agent harness, blueprint engine interleaving deterministic steps (lint, build, git) with AI reasoning, MCP tool layer exposing internal APIs, warm sandbox pool, and retrieval over the codebase, tickets, and historical PRs.
- Own the economics of the pipeline: observability over token usage, task latency, and cost per run; model-tier routing and cost-per-outcome gating.
- Embed governance into the pipeline — machine-generated audit trails, named ownership per feature, data-classification gates keeping production secrets and PII out of agent context.
Verification Engineering
- Engineer verification, not code: external behavioural scenario suites and holdout sets stored outside the codebase (agents never see the evaluation criteria), digital twins of the systems we integrate with, layered AI-reviews-AI gates, and automated remediation for self-healing services.
- Set and enforce the constraints that keep agent output correct on performance-critical paths: bulk ingestion throughput, large export streaming, memory behaviour, and complex SQL — validated by scenario and load gates rather than line-by-line reading.
Production Ownership & Collaboration
- Diagnose and resolve production data issues through agents — with deep enough functional understanding of every service you own to steer remediation quickly, in collaboration with Support and Engineering.
- Document the specs, scenarios, and pipeline decisions that govern the platform — artefacts both humans and agents can act on — and help other engineers make the same shift.
Skills & Experience
What you'll bring
- Demonstrated senior-level experience building production backend services — including .NET/C# and PostgreSQL systems — with a demonstrable shift from writing code to directing agents: the majority of your recent production output should be AI-generated under your direction.
- Hands-on mastery of AI coding agents — Claude Code strongly preferred — including parallel agent sessions, custom skills and sub-agents, and structuring codebases and context so agents work reliably.
- Excellence in specification writing — turning ambiguous business needs into precise, agent-executable PRPs and spec files. In this role, spec quality is the single most load-bearing skill.
- Strong verification engineering: behavioural and scenario-based testing, holdout evaluation, quality gates, and validation frameworks — the ability to trust and judge output you did not write.
- Strong architecture judgement: assessing whether an agent-proposed solution meets performance, scalability, and reliability requirements, and challenging designs that will not hold up under real workloads.
- Experience building agent infrastructure: MCP servers and tool layers, sandboxed execution, and CI/CD adapted for agent-driven workflows.
- Deep enough .NET/C# and SQL reading fluency to audit generated code when quality gates flag it — schema design, query plans, streaming vs. buffering, memory behaviour.
- Excellent written communication: specs, scenario definitions, and design docs that both humans and agents can act on.
Bonus points if you have
- Experience building digital twins or service virtualisations of external systems for agent evaluation environments.
- Experience with LLM evaluation at scale: eval suites, red-teaming AI-generated code, benchmarking agent workflows.
- Multi-tenant SaaS architecture experience, CDC/event-driven ingestion (Debezium, Kafka, RabbitMQ/MassTransit), or lakehouse exposure (Microsoft Fabric).
- Azure services, Docker, Kubernetes/Helm.