Production Support Engineer

What you will own:

First response and first-line resolution 

• Own the support queue end to end: acknowledge new requests promptly, classify them by severity, and drive every item to resolution or to a complete handoff. 

• Answer usage and configuration questions directly, from the documentation and your own understanding of the product. A large share of inbound volume is not a defect: there are questions, workflow misunderstandings, enhancement ideas, and reports that turn out to be working as designed. Resolving those yourself is one of the main contributions of the role. 

• Where the system behaviour is expected, walk the user through the correct workflow and close the loop with them. Where it is a defect, confirm that properly before escalating. 

• Go back to the requester for whatever is missing, including screenshots, record references, steps to reproduce, timestamps and environment detail. 

Technical investigation and triage 

• Carry out first-level technical triage: inspect application and cloud logs, run read-only database queries, reproduce the reported behaviour, and form your own view of the likely cause. 

• Read application code well enough to locate the relevant logic and confirm or rule out a hypothesis. 

• Work each ticket from more than one angle before drawing a conclusion, and recognise when you have reached the limit of what you can resolve on your own. 

• Treat our internal AI triage tooling as a starting point rather than an answer. Check its classification and its suggested cause, correct it where it is wrong, and add your own findings. 

Escalation and follow-through  :

• Escalate only the tickets that need it, with a reproducible case, logs, affected records, environment detail and your own analysis attached. 

• Create and own the tasks sent downstream to the engineering team, then follow them through fix, review and release, and verify in production. 

• Push back where a request sits outside agreed scope, and route it as an enhancement rather than letting it enter the queue as a defect. 

• Judge when to ask for help. Flagging early is preferred to spending a day stuck on the same problem. 

Client communication :

• Act as the named point of contact for client users, over email, the support portal and chat. You will be writing to experienced enterprise users, in some cases daily. 

• Keep them informed at an agreed cadence, including regular status updates on high-severity issues until they are resolved, whether or not there is progress to report. 

• Tell clients plainly what is known, what is being done, and when they will next hear from you. Tickets should not go quiet. 

• Hold your position courteously when a request is out of scope or a severity label is wrong, and know when to bring colleagues into the conversation. 

Knowledge, reporting and improvement  :

• Contribute to the knowledge base as you go, covering known issues, workarounds, runbooks and standard responses, so that a repeat occurrence takes less work than the first. The work you do here will contribute directly to the skills of the triage AI Agent.  

• Produce a weekly support report for each account, covering volume, mix, response performance, ageing, open high-severity items and the recurring issues that warrant a permanent fix. 

• Surface effort being spent outside agreed scope so that it can be reviewed. 

• Give feedback on the triage tooling. Where its classification or draft replies are wrong, explain why, so that it can be improved rather than worked around. 

What we are looking for 

Essential 

• Three to five years in application support, production support, technical support engineering, or a comparable client-facing technical role. 

• Python. Able to read and debug application code, and write small scripts to inspect data or verify behaviour. 

• Cloud platforms (AWS or equivalent). Comfortable navigating logs and monitoring to trace an issue through a running system. 

• SQL and PostgreSQL. Able to write your own queries to answer data-level questions without waiting for someone else. 

• A track record of resolving tickets rather than routing them. At interview we will ask you to talk through issues you closed yourself that others might have escalated. 

• Excellent written English. Most of your output is writing, and much of it goes directly to a client. You will also regularly be on calls with client users, being able to explain how the system works using vocabulary that is relevant to client, is important.  

• Self-starting and methodical. You take notes, follow up on open items without being prompted, and keep track of everything that is outstanding. 

• Comfortable being assertive. The role involves challenging a severity label, declining an out-of-scope request, and telling a client that what they are describing is not a defect. 

• Able to carry several accounts and unrelated products at the same time, and to switch between them without losing track. 

• Able to learn a complex domain application quickly. Our platforms carry substantial business logic, and supporting them well means understanding that logic rather than recognising screens. 

• Comfortable being accountable for response times and for the client relationship, not only for ticket counts. 

Desirable 

• Experience working to defined service levels in enterprise, regulated or audited environments. 

• Experience with an enterprise ticketing platform and a downstream engineering task tracker. 

• Additional languages. 

• Exposure to financial services, professional services, insurance, tax or transport operations. 

• Experience working alongside AI agents or automation within a support workflow. 

How success is measured 

Horizon 

What good looks like 

First 30 days 

You own the queue on your first account, acknowledge new requests within the agreed window, and classify severity correctly without help. Requests waiting on missing information are chased and moving. 

60 days 

Only the tickets that need an engineer reach engineering, and they arrive complete and reproduced. The weekly support report is running as a standing item, and the knowledge base covers the issues that recur. 

90 days 

The same model is running on a second account. You are proposing the product changes that would remove whole categories of ticket, and you own the support process itself, including intake, routing, escalation and client reporting. 

 The headline measure is the proportion of inbound tickets resolved without engineering involvement. 

Tooling 

Enterprise ticketing and task tracking, Microsoft 365, AWS (logs, monitoring and managed databases), PostgreSQL, Python, Git, and our internal AI triage tooling.