We are hiring a Lead Data Engineer who operates as a hands-on Solution Architect — a senior engineer who designs the architecture and writes the code, holds their own in live technical debate with a senior client-side architect, and runs discovery on ambiguous, partially-scoped briefs. This is the technical centre of gravity for a small delivery pod, not a detached oversight role.
The immediate placement is on an enterprise data-platform engagement leveraging Snowflake, dbt, and Azure DevOps, but this is explicitly a dual-lens hire. Xenoss is a boutique adtech services company that builds, scales, and rebuilds data platforms from scratch across rotating clients and stacks. The right person must satisfy the current project’s hard technical requirements AND carry the broader competency profile to lead future engagements, which skew predominantly adtech. Genuine adtech-platform depth is the single strongest signal for that future-lead potential.
Two capabilities are weighted especially heavily in this search, reflecting where we want more depth than the outgoing architect carried: * Deeper adtech background — programmatic, DSP/SSP/DMP/CDP, ad measurement, identity resolution; * Modern cloud data-engineering mastery on the Snowflake + dbt + Azure stack, specifically.
About the Project Client: a UK-based market research company. Its flagship product is a brand-tracking platform that measures consumer attitudes through monthly weighted surveys across 50+ metrics, including awareness, consideration, brand affinity, and NPS. The product is part of a broader analytics ecosystem, so everything built must be compatible with the wider platform rather than a single application.
Xenoss builds the Gold layer — the governance and packaging layer on Snowflake, sitting on top of the client’s existing Silver layer. The guiding principle is “build once, serve many”: one governed layer feeding dashboards, a commercial API, the Snowflake Marketplace, and direct enterprise data shares through consistent access controls.
Environment reality that shapes the role * Process-greenfield Snowflake estate at start — no CI/CD, tests, quality gates, or versioning, on a live business-critical dataset. The architect designs the productionisation foundations from scratch. * Mono-repo working model — Xenoss can open PRs but cannot merge. Savanta approval required (segregation of duties) * Two Snowflake accounts (UK + US) — multi-account topology to design around * Silver layer is a moving target, actively migrated in parallel — the architect designs against incomplete, shifting upstream inputs * Architecture-by-negotiation — most consequential decisions are co-owned with Savanta’s senior engineering architect, who owns ~70% of client-side technical dependencies
Key Responsibilities Solution Architecture (hands-on): * Own the end-to-end architecture of the Snowflake Gold layer — and personally build it, not just diagram it * Design governed consumption layers on the “build once, serve many” principle (One packaged layer feeding): * Client platform dashboards, * Snowflake Marketplace, * Direct shares * Make and defend consequential structural calls (e.g. dbt project topology, canonical identity model, dimensionality, lineage chain design) * Respect upstream source-of-truth boundaries — package and govern without recalculating business logic owned elsewhere * Future-proof the schema for known-future needs (identity bridging, clean-room) without over-building for them today
Snowflake & dbt Engineering: * Design and implement at multi-billion-row scale — Dynamic Tables, Streams + Tasks, clustering and materialisation strategy, p95 latency management * Build Row Access Policy (RAP) based multi-tenant entitlement, masking macros, classification, and suppression policies * Architect Secure Shares and Marketplace data products: versioned, governed, documented, with sample queries * Set dbt transformation strategy on the estate — mesh / multi-project structure, contracts, materialisation, tests, lineage, and data-model versioning for governed dataset contracts * Author Snowpark UDFs/procedures in Python where needed
Productionisation on an Immature Estate (Azure): * Stand up CI/CD for Snowflake on Azure DevOps from greenfield (not GitHub Actions — settled in client workshop) * Design for testability, versioning, and promotability — operate cleanly within no-merge PR governance and segregation-of-duties constraints * Establish quality gates, monitoring/alerting, and Snowflake cost governance where none existed * Design blue/green deployment for the governance layer
Discovery, Scoping & Client-Facing Architecture * Run discovery on vague or partially-scoped briefs — separate real scope from assumed scope and protect engagement margin * Hold your own in live technical argument with a senior client architect — concede cleanly when wrong, and ideally arrive independently at the same answer * Flex communication register across commercial (Head of Data Products), technical peer (client architect), delivery (PO), and analytics end-users * Treat documentation as a deliverable — PRD sections, dataset contracts, data dictionary, ADR-style rationale, reviewed by client engineers and product
Pod Leadership & Delivery * Act as the technical centre of gravity for a 3-FTE pod (architecture and delivery shape, not people-management) * Run 2-week sprint cadence — planning, demos, weekly status calls, phase-gate go/no-go reviews * Inherit and carry in-flight architectural decisions to close — capture reasoning, not just outcomes
Required Qualifications * 12+ years in software/data engineering, with a substantial track record in solution architecture AND continued hands-on delivery * Proven experience designing and building production data platforms at high scale (billions of rows; multi-billion-row table reasoning) * Demonstrated ownership of architecture in a client-facing or consulting context — has held an architecture line with senior client-side engineers * Experience standing up production discipline (CI/CD, testing, governance) on an immature or greenfield estate * English: strong written and spoken (C1+ effectively) — this is a client-facing, documentation-heavy, multi-register role
Technical Acumen (project-critical) * Snowflake — deep, hands-on: * Dynamic Tables, * Streams + Tasks, * Clustering/optimisation, * Performance at multi-billion-row scale, * Cost awareness * Row Access Policies (composition + performance), secure views/shares, multi-account topology * dbt — mesh / multi-project structure, contracts, materialisation, tests, lineage; able to set transformation-framework strategy on a fresh estate, not just operate an existing one * Azure — Azure DevOps for CI/CD pipelines (Snowflake deployment), and comfort across the wider Azure data ecosystem * Python — production data-engineering code; Snowpark UDFs/procedures; API services (FastAPI or equivalent: JWT auth, filtered exports, presigned URLs) * SQL — expert-level, including warehouse-scale query design and optimisation * Orchestration — Airflow (or equivalent) for pipeline scheduling and dependency management * Data modelling — canonical identity / master-data design, dimensional modelling, slowly-changing / versioned data, lineage design * Data governance — multi-tenant access control, masking, classification, suppression, commercial data-product packaging
Judgement & Soft Capabilities
The moments that matter in this role are scope calls and independent architectural convergence with the client — not “knows Snowflake.” Nice to Have Transferable / future-lead column — the tie-breaker that distinguishes someone who can run our next engagement from someone who can only fill this seat. Because Xenoss’s pipeline is predominantly adtech, genuine adtech depth is the strongest single signal here, not an optional extra. * AdTech platform depth — programmatic, DSP / SSP / DMP / CDP, ad measurement/attribution, high-volume event pipelines, identity resolution, clean rooms. (Weighted heavily — future-pipeline tie-breaker.) * Platform-agnostic cloud-DW depth — able to repeat the same reasoning on BigQuery / Databricks / Redshift if a future client differs * Data-product / DaaS commercial instinct — versioned data-product contracts, Marketplace listing packaging, Snowflake Native Apps * Database migration leadership — e.g. SQL Server -> Postgres at billion-row scale (schema mapping, parallel extraction, dual-write cutover) * Market-research / weighted-survey data familiarity — sample/suppression sensitivity (relevant to the current seat specifically) * MarTech / CRM platform background (Salesforce Marketing/Service Cloud) — adjacent to the adtech core * Pre-sales / engagement-origination experience — producing client-grade architecture artefacts and proposals
Ideal Candidate Profile
The right person is a hands-on architect — broad enough to own architecture, delivery shape, and client trust; deep enough to be unarguably credible to a senior client-side engineer. They have spent their career building real platforms that run in production at scale, and they still write code. They are not a slide-deck architect. They reason natively about multi-billion-row performance, can set a dbt strategy on a fresh estate, and know how to stand up CI/CD and governance on an immature Snowflake estate without breaking a live business-critical dataset. Snowflake, dbt, and Azure DevOps are genuinely in their hands — not aspirational lines on a CV. Critically, they bring judgement under ambiguity: they run discovery on a vague brief, separate real scope from assumed scope, and form opinionated architecture they can defend — and revise — in live debate with a senior client architect. They flex register naturally across commercial, technical, product, and analytics audiences, and they treat documentation as a first-class deliverable. The strongest candidates carry genuine adtech depth — they have architected high-volume programmatic, identity, or measurement data systems. For Xenoss specifically, that maps onto the future pipeline far more directly than a domain-neutral data architect of equal technical strength, and it is the clearest signal that this person can lead the next engagement, not just deliver this one.
