Instrument Risk Specialist Western Cape

You will build the systems that decide whether our hardest numbers are true - and then you will own that truth. This role owns the truth of the firm’s instrument, pricing and exposure data: the classification of every instrument, the validation of pricing and exposure, and the translation of regulatory and mandate intent into data that stands up to scrutiny. The role exists because, as a systematic investment manager, our value is created in the layer where raw market data becomes the classified instruments, valuations and exposures our clients and regulators rely on, and today much of that judgment lives informally, in spreadsheets and in people’s heads. This role turns it into governed, automated systems, and then remains accountable for what they produce. The shape of the role matters: it begins as a greenfield build (designing and standing up the tools, checks and classification logic) and matures into standing accountability: being the person who can say, and show, that these numbers are true, long after the systems that produce them run themselves. If your satisfaction ends the day a system ships, this is not your seat. The remit is deliberately narrow and deep: not the pipelines, not the platform, not the investment decisions, but the meaning and integrity of the numbers, and the systems that keep them honest. A small remit with real authority behind it.

Design and build the systems that classify instruments, roll credit ratings forward, and validate pricing and exposure across asset classes: automated, governed, and durable rather than one-off scripts, built on the firm’s cloud data platform (AWS, GitHub, Airflow orchestration) alongside the technology team. Translate regulatory and mandate intent into defensible data: interpret what a mandate, regulation or classification standard requires and encode it as auditable definitions and checks. Exercise the daily judgment the systems cannot hold: new-instrument classification calls, valuation edge cases, and the decision that a result is reasonable, not merely computed. Design the controls that catch silent failures (the plausible, confident, wrong number that no alarm flags) before they reach a client, an auditor or the board. Own the data contracts and vendor relationships for critical instrument and pricing sources, so a bad upstream number dies at the door rather than in a client report. Document the reasoning behind every system (assumptions, scope, what each check does and does not watch) so another capable person could repair it, not merely run it. Certify and stand behind the numbers when questioned, including to auditors, governance forums and clients: when a number is challenged, produce the evidence that it is right.

5–8 years in quantitative markets, risk or investment-data roles, at a level of owning outcomes rather than executing tasks. Demonstrated instrument-valuation and classification judgment across asset classes. Working fluency in credit and market risk and regulatory-reporting concepts (rating methodologies, capital and risk frameworks). Quantitative depth (derivative pricing, risk modelling) sufficient to judge whether a result is reasonable, not merely arithmetically correct. Strong programming: Python essential; SQL; others (C#, R, SAS) advantageous. A track record of building durable, well-structured data tools. Experience with modern data platforms advantageous: cloud warehouses or lakehouses (AWS), version-controlled delivery (Git), pipeline orchestration (e.g. Airflow), data-quality frameworks. Data-governance and stewardship experience: definitions, lineage, controls; caring that a number means the same thing everywhere it appears.

A postgraduate qualification in a quantitative discipline (quantitative risk management, financial engineering, statistics or similar), or equivalent depth demonstrated in practice.

Ownership and accountability. Builds things meant to last and to be owned, not handed off and forgotten; takes personal responsibility for a number being right years after the tool shipped. Sceptical rigour. Suspicious of their own automation; treats the quiet failure as the dangerous one and designs for it. Judgment under ambiguity. Comfortable making and defending the call where the rule runs out: a new instrument, an unusual structure, a contested interpretation. Legibility and communication. Explains reasoning as naturally as writing code; documents so that systems survive their author; communicates a technical judgment to a non-technical audience without loss. Business-first orientation. Starts from why the business needs a thing before reaching for a tool. AI-augmented engineering. Uses modern AI tooling as a genuine force multiplier while understanding, deeply, everything shipped with its help.