SDET 3 - Model Evaluation

Job highlights

• Design load and benchmark tests for inference services and batch pipelines
• Experience improving shared automation frameworks and raising team standards for maintainable tests.
• Build validation frameworks for computer vision models (preferably satellite or geospatial)
• Every release should come with a clear,data-backed go / no-go signal. . Make data quality a first-class gate

Job Summary

The Platform ML team deploys and validates every ML model behind AiDash's products. These models process satellite imagery, geospatial data, and weather data. As an SDET-3, you will own end-to-end quality for these models, from training data preparation through inference pipelines to performance at production scale. You will validate both functional correctness and scaling behavior, and you will use agentic AI tools to move faster without lowering engineering standards.

Responsibilities

• Own model quality before production. Build validation frameworks for computer vision models (preferably satellite or geospatial). Catch regressions in accuracy, precision/recall, and IoU before they reach customers.
• Build version-aware model regression. Set up pipelines that compare inference outputs across model versions on curated golden datasets. Every release should come with a clear, data-backed go/no-go signal.
• Make data quality a first-class gate. Validate training and evaluation datasets for label quality, class balance, geographic and seasonal coverage, leakage, and drift. Ensure models are tested on data that reflects real-world conditions.
• Validate inference at scale. Design load and benchmark tests for inference services and batch pipelines. Find the throughput, latency, and cost limits on AWS and Kubernetes before customers do.
• Harden the platform end-to-end. Automate API, contract, and data-pipeline tests across distributed services. Use logs, metrics, and traces to speed up root-cause analysis.
• Raise the bar for the team. Strengthen the shared automation frameworks and set standards for maintainable, non-flaky tests. Show how tools like Cursor and Claude can speed up test authoring, triage, and reporting while staying under human review.
• Influence release decisions. This is a high-visibility role. Your quality signals will directly shape what ships across AiDash's product lines.

What we're looking for

We are looking for a senior QA engineer with deep backend automation experience who can validate ML models and backend systems end-to-end, and who actively uses agentic AI development tools (e.g., Cursor, Claude) to accelerate test design, automation, debugging, and reporting - without compromising fundamentals, coverage, or engineering quality.

• 8-10 years in software testing and test automation, with strong ownership of backend system quality
• Hands-on backend automation experience using at least one of:
• Java + RestAssured
• Or Python + pytest
• Experience testing ML applications (model/service behavior, data quality, inference pipelines, evaluation/regression of model-backed features)
• Broader stack familiarity: observability (logs/metrics/traces), cloud (AWS), contract testing, data-pipeline validation
• Experience improving shared automation frameworks and raising team standards for maintainable tests
• Solid grasp of testing fundamentals and best practices (test design, risk-based coverage, isolation, maintainability, flaky-test control, clear defect/RCA discipline)
• Strong coding ability in Java or Python (proficient in at least one), with willingness to learn the other (or additional languages) as needed
• Good understanding of backend system design and how services interact in distributed systems
• Practical exposure to databases - SQL and/or NoSQL (at least one is mandatory)
• Working knowledge of application deployment and underlying orchestration (e.g., containers, Kubernetes, service rollout/health concepts)
• CI/CD pipelines - must have (Jenkins/GitHub Actions/Bitbucket Pipelines or equivalent): triggering suites, reading failures, improving pipeline reliability
• Performance testing - must have (JMeter/k6/Locust or equivalent): load/benchmark basics, interpreting results, flagging SLA/risk
• Exposure to agentic development tools such as Cursor and Claude (or similar) for coding, test authoring, log/triage assistance, and documentation - used with human review and quality gates

Disclaimer: This job posting has been aggregated from external source. Role details, content, and availability are subject to change. Applicants are advised to confirm the latest information directly on the company website before applying.