Product Engineer / Forward Deployed Engineer

Function: Auraa Engineering • Level: Senior or Lead • Product: Auraa by Covasant, built on Databricks

Help build Auraa, our agentic data platform on Databricks, and see it work on real customer data through one-week Proofs of Value.

Building the foundation for enterprise AI

Covasant is an AI native company building autonomous, agent driven systems for the up-to-date data and AI stack. Auraa, our agentic data activation platform, builds governed lakehouses on Databricks, giving data and AI leaders the capacity to move from infrastructure to AI outcomes, faster.

We are ISO/IEC 42001 certified, and Google Cloud selected us as one of few AI-native partners worldwide for its Gemini Enterprise Transformation program. Our customers include financial services firms, insurers, manufacturers and retailers.

The fastest path to AI-ready data on Databricks

We launched Auraa at Databricks Data + AI Summit 2026. It is our agentic data activation platform, built natively on Databricks. Customers describe what they need in plain language, and Auraa's agents discover sources, build pipelines, enforce data quality and register everything in Unity Catalog.

Auraa runs on governed, versioned metadata rather than generated code, so every decision is auditable and every result reproducible. We take it to market together with Databricks.

Build the product, prove it with customers

Most of your time goes into building Auraa with the Auraa Engineering team. From time to time you take it to a customer for a one-week Proof of Value (PoV) on their live data, working alongside a Solution Architect who scopes it with you up front.

What you learn in those weeks comes back into the product,



so your design choices get tested against real systems quickly.

What you will build

- Ingestion and delivery. Extend autonomous source-to-Silver delivery across relational, SAP, file, object storage and streaming sources, with schema evolution, incremental loads and error recovery.
- Governed metadata. Design the specifications and registries behind pipelines and quality rules, so the same specification always produces the same result.
- Lakeflow orchestration. Generate Databricks Asset Bundles and Lakeflow Declarative Pipelines in code.
- Quality and self-healing. Profiling, LLM-assisted rule proposals and drift detection, plus automated recovery from schema drift, late data and failed connections.
- Multi-tenancy and governance. Unity Catalog isolation, automated grants, RBAC and identity integration that fails closed.

How a Proof of Value runs

1. Discovery. Before the week starts, review the customer's workspace, Unity Catalog setup, network path and identity model with the Solution Architect.

2. The build. Connect the source, run ingestion, review proposed quality rules with the customer's data steward, then promote to production.

3. The readout. Present results to the customer's leadership on the final day, including what did not work.

4. Findings. Write up what you learned and turn it into product backlog.

What you bring

- 10 to 15 years in data engineering or data platforms,



including senior or architect roles where you owned the design.
- Deep Databricks. Spark, PySpark, Delta Lake, medallion architecture, Unity Catalog, Asset Bundles or Workflows, and a feel for compute costs.
- Python as a software engineer. Clean abstractions, packaging and tests you wrote yourself.
- Production ETL and ELT at scale. Incremental and full loads, SCD handling, schema evolution, idempotency and replay.
- Framework thinking. You have built a reusable engine that other teams ran on.
- Governance and CI/CD in practice. Catalogs, lineage and access policy in production, with Git-based promotion and infrastructure as code.
- Clear communication. You can explain a design to engineers and executives alike, and you write decisions down.

What will stand out

- Applied GenAI or agentic systems in a data engineering context, including MCP or tool calling.
- SAP or streaming sources (Structured Streaming, Kafka, change data feeds).
- Consulting, pre-sales or SI delivery experience.
- Databricks Data Engineer Professional or Generative AI Engineer Associate certification.

How we hire

No whiteboard puzzles. Each conversation reflects real work.

1. Code and design conversation. Walk us through something you built and the trade-offs you made.

2. Practical exercise. Python and Spark against a deliberately messy source.

3. Design discussion. Multi-tenancy, idempotency or failure containment.

4. Field scenario. A source schema changes mid-PoV. Talk us through your approach.

With your application, share something you built and the hardest failure mode you have designed around.

Covasant is an equal opportunity employer. We hire on capability and judgment.

📌 Product Engineer / Forward Deployed Engineer (India)
🏢 Covasant Technologies
📍 India