AWS Data Engineer for ELT Pipelines Design

Job ID: 40145151

Budget: $25 – $50 USD

Summary

Overview
We’re looking for an experienced AWS Data Engineer to help us design and implement production-grade ELT pipelines on AWS. The objective is a scalable, well-governed data foundation that supports accurate reporting and analytics with clear lineage, repeatable transformations, and reliable orchestration.

This is not a one-off scripting project. We want an engineer who can make sound architectural decisions, implement infrastructure and pipelines cleanly, and create documentation we can reuse across clients and datasets.

Current AWS Stack
You will work primarily with:

S3 (data lake storage)

AWS Glue (crawlers, jobs, Data Catalog)

Athena (querying)

Amplify (light frontend serving curated inputs)

What We Need Built
We need help defining and implementing the end-to-end flow:

Architecture & Data Modeling

Propose a practical Bronze/Silver/Gold design for our use case

Establish conventions: folder structure, partitioning strategy, naming, table registration, and schemas

Ingestion (Raw/Landing)

Build repeatable ingestion patterns into S3

Set up Glue Data Catalog assets and crawlers as appropriate

Ensure ingestion is auditable (run logs, basic data quality checks, and schema change approach)

Transformations

Implement transformations (Glue / Spark / SQL) into curated tables

Produce analytics-ready datasets with agreed-upon business logic

Optimize for performance: partitions, file sizing, and query cost control

Orchestration / Reliability

Implement a reliable scheduling/orchestration approach (recommend best fit: Glue workflows, Step Functions, MWAA/Airflow, etc.)

Add a monitoring/alerting strategy appropriate for a small team (CloudWatch-based is fine)

Governance & Access

Configure permissions and table governance properly

Maintain clear lineage from raw to curated

Handoff

Provide clear documentation: architecture diagram, runbook, and how to extend the pipeline

Ideal Candidate (Must Have)

Strong hands-on experience with S3, Athena, Glue, Python

Experience building production ELT/ETL pipelines (not just ad hoc scripts)

Solid understanding of data design (partitions, Parquet, table formats, cost/performance)

Comfort with SQL and data modeling for analytics-ready datasets

Ability to communicate tradeoffs clearly and propose a clean architecture

Clear updates, short design review upfront, then implementation sprints

After reviewing this proposal, I will proceed with a 4-hour test. The detailed scope is outlined below.

Architecture + Data Modeling Design Review

Project Summary
Create data pipelines for marketing and sales data that will feed Tableau. You will ingest:

HubSpot (sales and lead opportunities, revenue tracking for closed deals)

Google Ads (marketing campaigns, spend)

Manual inputs for goal setting and commission setting

Propose the best design to complete this work. The work must take place in an AWS environment, with Athena preferred as the database solution.

Deliverable: Written Design Doc
Please address the following:

Assumptions & Open Questions

Assumptions about volumes, refresh cadence, number of clients, environments (dev/prod), etc.

Proposed Architecture

High-level diagram (simple boxes/arrows is fine)

Services used and why

Data flow overview

Orchestration + Reliability

Tool choice and reasoning

Dependencies, retries, and idempotency approach

Backfill strategy (re-run date ranges safely)

Monitoring + Data Quality

Minimum viable checks (row count deltas, null rates, duplicates)

Alert triggers and definition of “pipeline failure”

Cost/Performance Considerations

Athena scan minimization strategy

Partitioning vs partition projection (if relevant)

Small file mitigation

Handoff / Runbook Outline

What gets documented and how a new engineer would extend the pipeline

Pricing Plan & Summary Scope
a. Milestones including timelines, acceptance criteria, and estimated effort (hours)
b. Estimated maintenance and operating costs

Pros and Cons of Designs

Possible upgrades to the system

Please let me know your proposed approach.