Senior Data Engineer

Reports to - Engagement Lead

Shift - Night shift, 9:00 AM - 6:00 PM US Eastern)

Role Summary

We are looking for a senior data engineer who has built and run a data platform end to end, and who will own one for our US clients during their business day. You work across the whole stack: ingestion from source systems and APIs, layered dbt models in BigQuery, orchestration in Airflow, and the BI layer that business users actually look at. When a number looks wrong you trace it from the dashboard back to the source and fix it at the cause. When the platform is healthy you build: new sources, new models, new metrics, faster queries, better tests. You work directly with the client's analysts and business leads, including on live calls, and you stay accountable until the outcome is delivered and the business trusts it.

Key Responsibilities

The role is accountable for the following outcomes. Each bullet starts with the outcome owned, followed by the supporting activity.

• Own the platform end to end, in US hours - Be the senior engineer responsible for the full data flow, from source to dashboard, while the client is working. You decide what to build, fix or improve next, keep one clear owner on every piece of work until it is done, and hand over cleanly to the India team at the end of your shift and pick up cleanly at the start.

• Build and run ingestion - Design, build and maintain batch and incremental pipelines from operational systems, marketplaces, advertising platforms and APIs, using managed connectors where they fit and custom Python where they do not. Handle schema drift, latearriving data, rate limits and backfills deliberately, and make every load idempotent and rerunnable.

• Own the dbt transformation layer - Build and evolve the dbt project across raw, staging, master and presentation layers: incremental strategies, snapshots, tests, model contracts, macros and documentation. Business logic is defined once in the model layer and reused everywhere rather than re-implemented in each report.

• Run orchestration as an engineered system - Own the Airflow (or equivalent) DAGs: dependencies, retries, backfill behaviour, alerting and run history. Diagnose failures from logs, lineage and run metadata, fix the cause rather than the symptom, and prevent repeats with a test, an alert or a code change.

• Own the BI and consumption layer - Build and maintain the semantic and reporting layer in the client's BI tool, such as Power BI, Tableau, Looker or an embedded analytics tool. Be able to trace any dashboard number back to its source table and its transformation logic, and reconcile it against the source system of record before you tell anyone it is right.
• Work directly with US stakeholders - Be the voice of Dunboxed to the client's analysts and business leads during their day. Run requirement conversations, join live calls and debug with them in real time, explain root causes in business language, and give honest timelines. Raise design, scope or commercial questions early, with a complete write-up covering symptom, impact, evidence, options and recommendation, so nobody has to re-ask the client.

• Ship safely on a shared codebase - Deliver changes through tests, code review, CI/CD and pre-production validation, with a hotfix path for the genuinely urgent. Check the blast radius before you merge so a fix for one client or model never breaks another. Destructive operations (drops, truncates, overwrites, full refreshes) are flagged and reviewed before they run.

• Make data quality, performance and cost visible - Put freshness, volume, schema and referential checks in place with alerts that fire before the business notices. Tune slow or costly BigQuery workloads (partitioning, clustering, query plans), and apply access control and PII handling from the start.

• Build with agentic coding, and keep the standard high - Use Claude Code or an equivalent agentic coding tool daily to scaffold models, write tests, investigate failures and produce documentation, within guardrails. Review every change as if a junior engineer wrote it, and never delegate business definitions or destructive operations to an agent unchecked.

• Document, mentor and build patterns - Keep per-client context, runbooks and handover notes current so any engineer can pick up any account. Review pull requests, coach newer engineers, and turn repeated work into reusable accelerators (macros, project templates, pipeline patterns).

Required Qualifications

Candidates must demonstrate all of the following. Items here are non-negotiable - if a candidate is missing one, they are not a fit.

• Experience - 6-10 years of hands-on data engineering, including production ownership of at least one platform you took from raw sources through to dashboards that business users rely on. You have built the stack, not only maintained pieces of it.
• SQL and a cloud warehouse - Expert SQL (window functions, incremental logic, query plans) and production experience on BigQuery, or a comparable warehouse such as Snowflake, Redshift or Synapse with a clear route to BigQuery. Partitioning, clustering, cost control and access control.
• dbt in production - Hands-on dbt (Core or Cloud) on a real project: layered models, incremental models, snapshots, tests, macros, packages, environments, documentation, and CI that runs only what changed. You can explain how you structure a project and why.

• Python - Strong Python for ingestion, API integration, automation and data-quality tooling, and the ability to read and fix someone else's code without breaking it.
• Ingestion - Batch and incremental ingestion from databases, SaaS and marketplace APIs and files, using managed connector tools (Fivetran, Airbyte or similar) and custom pipelines. You know how to detect and recover from late, duplicated, missing or changed source data.
• Orchestration - Production Airflow (including Cloud Composer) or a comparable orchestrator such as Dagster, Prefect or ADF: dependency design, retries, backfills, monitoring and debugging a failed run from its logs.
• BI and metric tracing - Hands-on delivery in at least one BI tool (Power BI, Tableau, Looker or similar), including modelling for it, and the discipline to reconcile a dashboard metric to its source before reporting it.
• Engineering discipline - Git and pull-request workflow, code review, testing and CI/CD, and the judgment to ship a fix to a shared codebase without breaking other models or clients.

• Production ownership - Habits of triage by severity and impact, honest status updates, rootcause analysis and handover across time zones. You do not hand a problem on without a complete write-up.
• AI-assisted engineering - Regular use of Claude Code, Cursor, GitHub Copilot or comparable agentic tools on production work, with a clear view of what you delegate to them and what you deliberately do not.
• Client communication - Clear, confident English, written and spoken. You can join a live call with a client, debug with them, explain the cause in business language, and stay calm when an issue is escalated.
• US-hours night shift - Able to work 9:00 AM - 6:00 PM US Eastern Time from Hyderabad on a sustained basis. That is roughly 7:30 PM - 4:30 AM IST (6:30 PM - 3:30 AM IST during US daylight saving time). You work with limited backup during the shift and learn a new client's data quickly.

• Education - An engineering degree (B.E. / B.Tech or equivalent) from a reputed institution.
• Right to work and location - Eligibility to work in India without sponsorship, and based in or willing to relocate to Hyderabad.

Preferred Qualifications

Items here strengthen a candidacy but are not deal-breakers.

• US client experience - Prior experience working US hours or directly with US clients, and comfort with their expectations on responsiveness and communication.
• eCommerce and omnichannel data - Fluency in how contribution margin, customer cohorts, attribution, inventory and marketplace metrics are really calculated, and where source data misleads.
• Semantic or metrics layer - A dbt semantic layer, LookML or another governed metrics layer, and experience with embedded analytics tools.
• Data-ingestion products - Experience building or operating connector-based ingestion products similar to Fivetran or Airbyte, including connector failure modes.
• BigQuery administration - IAM and row-level security, slot and cost management, and budgets with alerts.
• Observability and incident tooling - Elementary, Soda, Great Expectations or a commercial observability tool, and ticketing or incident-management tooling in an SLA-driven environment.
• Mentoring - Experience coaching other engineers and setting standards for code review, testing and documentation.
• Certifications - GCP Professional Data Engineer or dbt Analytics Engineering. A signal, not a substitute for shipped work.
• Degree - A degree in Computer Science, AI, Data Analytics or Data Science.