19 Nov
|
Saras Analytics
|
Hyderabad
19 Nov
Saras Analytics
Hyderabad
Senior Data Engineer, DBT & BigQuery
You'll be the hands-on engineer behind our dbt, BigQuery, and Git codebase — building and improving the models, pipelines, and data-quality checks every client runs on — and you'll do it with Saras' internal AI agents handling the first pass. You review what the agents produce, own what reaches production, and make the agents better with every build, so one senior engineer delivers what used to take a team.
The Role
You'll work primarily in dbt, BigQuery, and Git, reporting to the Senior Data Engineering Lead. You design, build, and maintain the staging, master, and presentation datasets and the shared dbt codebase that the product team build on, and you do most of that work through Saras' internal AI agents: you set them up with the right context and conventions, direct them at the task, review every line they produce, and own the result. You are hands-on, not a manager: you write and review code every day. You are the technical escalation point for dbt and BigQuery problems that go beyond one account. You work alongside the FDEs who own client solutions, the US support team that resolves customer requests, the data-ingestion team that owns connectors, and the Product Manager who owns the roadmap.
You own whether the code is right: correct, tested, performant, cost-productive, and maintainable.
The Bar
Correctness first, then speed. AI agents make output fast, so the bar falls on the engineer: you understand and can defend every line that reaches production, whoever or whatever wrote it. Nothing merges without tests, review, and a pass through pre-prod, and numbers are reconciled against the source of truth before a client sees them.
You keep BigQuery cost in view: partitioning, clustering, incremental logic, and slot use are designed and measured, not discovered on the bill. Branches are small, history is clean, and commits explain why. You surface risks early and close every issue with a root cause and a fix.
The lanes are explicit: you own the dbt and BigQuery code and its standards; the FDE owns the client solution; the US support team owns request resolution in US hours; the Account Manager owns the relationship and the commercial outcomes; the Product Manager owns product onboarding, enablement, and roadmap.
What You'll Own
1. Own the dbt codebase. Build and maintain the shared dbt models, macros, packages, and tests (EDM-Insights) across the staging, master, and presentation layers. Keep client-specific logic configurable rather than forked, retire technical debt such as SELECT * and hard-coded variables, and make every change safe for every client that uses the code.
2. Build with the internal AI agents. Make Saras' AI agents your default way of working. Give them the context, conventions, and acceptance criteria, have them draft models, tests, documentation,
and fixes, then review and validate what comes back. Never ship what you haven't understood.
3. Make the agents better. Capture what you learn: gaps in agent context, repeated mistakes, new conventions, and prompts and skills that work. Feed it back to the team that maintains the agents, so each build makes the next one faster and more correct.
4. Own BigQuery design and performance. Design partitioning, clustering, incremental and materialisation strategies, and tune slow queries. Track slot and on-demand spend, reconcile it against billing, and find and fix the expensive models.
5. Own BigQuery access and security. Manage datasets, IAM, PII masking, authorized views and datasets, and metadata protection as code, so each client's data stays isolated and secure.
6. Own the Git workflow and releases. Set the branching strategy, pull-request standards, code-review practice, and CI checks, and run the path from pull request to pre-prod to production, with test DAGs on representative clients. Review pull requests promptly, keep hotfixes inside the same process, and never push directly to main.
7. Own data quality. Build and run the data-catalog and data-quality checks — completeness, freshness, reconciliation, and anomalies — on every build. Make checks configurable so FDEs can add them without new code. Resolve what they flag, write the root cause, and fix the class of problem, not only the instance.
8. Be the escalation point for dbt and BigQuery. Take the problems that go beyond one account: complex logic, platform limits, performance, and cost. Respond within the day, own the fix to closure, and turn each repeat problem into a macro, a check, documentation, or an agent improvement.
9. Set standards and mentor. Write and maintain the modelling conventions and templates. Mentor FDEs and data engineers on dbt, SQL, Git, and working with agents, and take part in hiring for data-engineering roles.
10. Run your delivery hygiene and partner across the lanes. Keep the data dictionary, lineage, documentation, task tracker, and timesheets current. Work with FDEs, the US support team, and the data-ingestion team to unblock delivery, and pass patterns you see to the Product Manager as roadmap input.
What Winning Looks Like First 90 days: you know the shared dbt codebase and the BigQuery setup in depth, you work through the internal AI agents by default, and you've shipped at least one meaningful improvement end to end,
such as a reusable macro, a cost reduction, or a data-quality check. Your pull requests need little rework.
Steady state: you ship far more than one engineer could without agents, and reviews still hold no surprises. Changes pass pre-prod cleanly, BigQuery spend is explained and trending down for the same workload, and the agents improve because of what you feed back. Your Lead trusts what you ship without re-checking it, your teammates enjoy working with you, and you're ready to lead a larger platform area.
Who Thrives Here
Any of these backgrounds can get you in the room. Across all of them, you bring 6+ years (typically 6–10) of hands-on data engineering work, at least 3 of them in dbt and BigQuery (or a comparable warehouse).
- You've been an analytics or data engineer who built layered dbt models — raw, staging, masters, presentation — and are ready to own the shared codebase they live in.
- You've been a software engineer who moved into data, and are ready to own the warehouse and transformation layer end to end.
- You've built dbt projects for many clients at an analytics consulting or services firm, and know what makes a codebase reusable.
Whichever route you come from, you have:
- A degree from a reputed engineering college.
- Expert dbt: macros, packages, incremental models, tests, documentation, and one project serving many clients.
- Expert SQL and BigQuery: query performance, partitioning, clustering, and cost control.
- Strong Git: branching, rebasing, conflict resolution, clean history, code review, and CI/CD.
- Strong data-modelling fundamentals across the raw → staging → master → presentation layers.
- Working Python, enough to script, test, and automate around dbt.
- Daily use of AI coding tools and agents such as Cursor, Claude, or GitHub Copilot, and the judgement to verify their output: you can tell when an agent is confidently wrong.
- Production debugging and root-cause analysis, and experience building data-quality and testing checks.
- Clear written communication: pull-request descriptions, decision notes, and documentation others can follow.
It's an even stronger fit if you also bring:
- An engineering degree in Computer Science, AI, Data Analytics, or Data Science.
- Ecommerce data fluency: you know how contribution margin, customer cohorts, attribution, and inventory are really calculated — and where the source data misleads.
- BigQuery access control and infrastructure as code (Terraform or similar).
- Experience with orchestration tools such as Airflow.
- Experience building or tuning AI agents, prompts, skills, or evaluations for engineering workflows.
- Experience with multi-tenant data platforms, where one codebase serves many clients.
- Experience mentoring other engineers.
📌 Senior Data Engineer (Hyderabad)
🏢 Saras Analytics
📍 Hyderabad