Analytics Engineer (Reporting & AI)

Full-time Cape Town (Remote)

About Us

Orion Labs is a small, fast-growing software development and DevOps consultancy based in Cape Town, South Africa. We work on modern cloud, analytics, platform engineering, and AI-related projects for local and international clients. We’re collaborative and hands-on with dynamic technologies. You’ll join a supportive team with real ownership, lots of room to learn, and clear opportunities to grow your career while doing meaningful work for great clients.

Job Description

We’re hiring an Analytics Engineer to work on a data platform we’re building for a client in the customer support industry.

One part of the job is Metabase: daily operational dashboards, splits by channel, queue, brand, and campaign, and whatever a manager asks for that week. When someone wants a new metric, you’re the person they ask.

The other is quality assurance. Only a small sample of tickets gets reviewed by hand today. You’ll build a process that uses LLMs to score tickets against the client’s quality rubric so they can cover all of them.

Hands-on, full ownership, direct access to leadership. There’s no committee between you and shipping a change.

Responsibilities:

Build and maintain the Metabase dashboards that operations, workforce management, QA, and campaign managers rely on.
Turn loosely worded requests into agreed metric definitions, then ship them.
Extend the dbt models behind those dashboards, with tests and documentation.
Build the LLM scoring pipeline: prompts from the rubric, model choice, and running it across full ticket volume at a sensible cost.
Check the scores against human reviewers until people trust them, then build the reporting so leads can see results by agent, queue, brand, and channel.
Own data quality. When a number looks wrong, you’re the one who finds out why.

Qualifications:

Bachelor’s degree in Statistics, Computer Science, Information Systems, or a related field.
Strong SQL. Joins, window functions, CTEs, and the judgement to know when a query needs rewriting.
Solid dbt experience: models, tests, and documentation.
Metabase, or another BI tool (Power BI, Looker, Superset, Tableau) and the appetite to switch.
Postgres: schema design, indexing, and reading a query plan.
You’ve used LLMs for real work, not just chat. You’ve written prompts that had to hold up across thousands of inputs and worked out whether the output was any good.
Git, and clear writing. You’ll document a lot and explain numbers to people who don’t work with data.

Nice to Have:

Python for scripting and ad-hoc work.
Contact centre or BPO background, so AHT, FRT, CSAT, and QA scoring aren’t new to you.
Zendesk, Intercom, Klaus, Aircall, or similar helpdesk and telephony reporting.

Technical Stack:

Warehouse: PostgreSQL
Storage: Azure Data Lake Storage Gen2
Ingestion and orchestration: Azure Data Factory
Transformation: dbt (SQL-based modelling, version controlled in Git)
Visualisation: Metabase
Quality scoring: LLM-based evaluation over ticket and conversation data
Source systems: helpdesk, telephony, and CRM platforms covering phone, email, and chat

Benefits

Competitive pay, a learning and certification budget, and a small team that values balance and doesn’t do ego. You’ll own the reporting and QA work outright, with a direct line to leadership and full Azure access. Cape Town based and remote, with the occasional get-together in person.

Inclusive Hiring

We hire for skill, potential, and values. In South Africa’s diverse context, we welcome applicants from all backgrounds and aim to provide a fair, inclusive process. If you have the skills and drive, we’d love to hear from you.