Data Scientist 2

<p>Data Scientist - 2 (DS-2) </p><p> Job family: Data Science<br /> Level: DS-2 (equivalent to MLE-2 / AIE-2)<br /> Scope of impact: Feature<br /> Theme: Grows and Acts — completes scoped modelling tasks and<br /> improves team process<br /> <br /> Why this role exists </p> <p> Product outcomes need deep problem ownership and tight iteration<br /> with PMs and product engineering. A DS-2 turns a scoped product<br /> problem into a calibrated model or decision system, ships it<br /> through the standard production path, and owns its performance<br /> after launch. You operate with minimal guidance on a defined<br /> feature, not the whole domain. </p> <p> What you own<br /> <br /> - A scoped modelling problem framed as a DS task: hypothesis,<br /> success metric, offline and online evaluation plan.<br /> - Calibrated predictive or causal models with well-behaved<br /> probabilities and effect estimates.<br /> - Repeatable pipelines integrated with production workflows, not<br /> one-off notebooks.<br /> - Basic model monitoring for the features you ship.<br /> - Post-launch performance of your model and its link to the<br /> target KPI; iterate using telemetry.<br /> <br /> What you do not own (yet) </p> <p> - Platform uptime and shared serving infrastructure (ML<br /> Engineering owns this).<br /> - Domain-wide priority setting across multiple initiatives (DS-3<br /> and above). </p> <p> What you'll do (proficiency expectations at L2) </p> <p> Data-driven decision making<br /> - Build calibrated predictive or causal models with sound<br /> probability and effect estimates.<br /> - Articulate the impact of uncertainty and select an applicable<br /> course of action with minimal guidance.<br /> - Stress-test findings with simple mental models or simulations<br /> before trusting them. </p> <p> Technical expertise<br /> - Set up fully reproducible environments for your own work and<br /> share the guides with peers.<br /> - Package work into repeatable pipelines and integrate them with<br /> production workflows.<br /> - Implement basic model monitoring. </p> <p> Applied ML/AI/DS<br /> - Frame and scope an opportunity as a DS problem, and pick the<br /> right solution family (prediction, optimization, causal).<br /> - Review recent literature, build reproducible pipelines, and<br /> fairly compare alternative models.<br /> - Run controlled pilots that connect model uplift to a target<br /> KPI.<br /> <br /> Experimentation and inference<br /> - Frame a testable hypothesis and pick the right design (A/B or<br /> hold-out).<br /> - Run multi-metric or stratified tests with power checks and<br /> CUPED variance reduction.<br /> - Conclude using confidence intervals, state the limitations,<br /> and tie results back to a target KPI.<br /> <br /> Strategy and influence<br /> - Scope an opportunity into a well-posed DS problem, naming the<br /> RoI and the product and process changes it implies.<br /> - Align stakeholders on the KPI leverage of a proposed approach<br /> and secure agreement on scope and goals.<br /> - Coordinate with engineering and product leads to launch<br /> features where the model provides core value; shape planning and<br /> risk assessment.<br /> <br /> How you work with others </p> <p> - PM: co-own the outcome and prioritisation for your feature.<br /> - Product Engineering: integrate your model into customer-facing<br /> experiences.<br /> - ML Engineering / AI Engineering: consume platform primitives;<br /> collaborate on evaluation, reliability gates, and production<br /> readiness. </p> <p> What we expect from a strong DS-2<br /> <br /> - Ships production artifacts on the standard path, not<br /> prototypes that stall at the production boundary.<br /> - Improves at least one team process (templates, reviews,<br /> reproducibility) beyond their own tasks.<br /> - Owns outcome integrity: model outcomes stay aligned with<br /> product outcomes after launch. </p>