AI Full-Stack Developer, Maintenance & Enhancements

Hiring for PULS Group — via Flexiple, the recruiting partner on this role.
About the CompanyPULS is a German manufacturer of industrial DIN-rail power supplies, used by automation OEMs worldwide. Headquartered in Munich with about 1,700 people globally, PULS is building an India-based IT and AI team that works directly with its Munich Software Development department. The team is relaunching the global PULS website as an AI-ready platform for 21 country sites in 7 languages and runs a growing portfolio of in-house AI solutions on Azure.
About the RolePULS runs live AI solutions used every day by Purchasing, Finance, Logistics, R&D and Sales: chat agents integrated with SAP, scheduled Azure jobs, document pipelines, workflow apps and dashboards. You will keep them reliable and keep improving them. That means resolving incidents at the root cause, shipping change requests in regular maintenance releases, and keeping everything monitored and documented. You will work from India with core hours overlapping the Munich morning (about 13:00–18:00 IST), with no night shifts and no solo on-call, alongside the AI team, DevOps and Cybersecurity. Onboarding is in Munich, with one or two visits a year after that.
What you'll doOwn the incident and change-request queue for all live AI solutions: reproduce on the test stage, separate bugs from change requests and agree weekly priorities with the AI Team LeadFix root causes across the stack: TypeScript/React frontend, Node.js or Python (FastAPI) backend, scheduled jobs, SAP OData integration, PDF and mailbox automation, and write the test that proves each fixShip enhancements in maintenance releases (new SAP fields, document layouts, extended agent skills) and turn larger requests into specifications for the project teamUse Claude Code and Codex to draft fixes and tests, then review, verify on the preview stage and mergeCatch rising error rates, silently failed jobs and token-cost drift with KQL in Log Analytics; release prompt, skill and model updates on Azure OpenAI only after regression cases and an LLM-as-judge review passRun Azure Container Apps and jobs across preview, pilot and production with DevOps: Bicep, rollbacks, Key Vault secrets, Entra app registrations and Teams app updatesTake over solutions built by business teams and keep runbooks and documentation current
What you bring4–6 years of full-stack development on production systems you not only built but kept runningTypeScript/React and Node.js, plus Python with FastAPI; SQL with PostgreSQL or Cosmos DBAzure in daily practice: Container Apps, Cosmos DB, PostgreSQL, Key Vault, Service Bus, Entra ID, Bicep and the Azure CLIHas shipped LLM applications to production: tool calling, prompt guardrails, RAG (pgvector) and evaluationReal debugging depth, e.g. has traced a 401/403 through an OAuth on-behalf-of chain and an API gateway to its root cause, and reads logs with KQLAt least 6 months of daily Claude Code or Codex use on production code. Expect live pairing on a real ticket in the interviewSecrets management, RBAC and GDPR as everyday practice; clear written English (C1) for non-technical colleagues in Jira and docs
Nice to haveSAP OData integration experiencePowerShell for Windows-based deployment and operations toolingMicrosoft Teams app development
Originally posted on: 29 Oct 2026

Skills
Python, JavaScript, TypeScript, React, Node.js, FastAPI, Azure, SQL, PostgreSQL, LLM

Work arrangement
No