Anirban B.

Hello, I am Anirban - a Lead SDET and AI Evaluation Engineer with 13 years of experience designing automation, validation, and quality frameworks for AI systems, distributed data platforms, and cloud-native applications. My current focus is evaluating generative and agentic AI systems. I help teams move beyond subjective “looks good” testing by establishing structured golden datasets, measurable rubrics, repeatable evaluation pipelines, and evidence-based quality gates. Agentic AI and LLM Evaluation * Designing a multi-agent evaluation framework for agents built using Google ADK * Reviewing golden-dataset preparation, expected outputs, and evaluation coverage * Creating weighted rubrics and scoring criteria for complex AI-generated deliverables * Evaluating factual accuracy, completeness, consistency, groundedness, reasoning quality, and instruction adherence * Identifying edge cases, hallucinations, workflow failures, and unsupported conclusions * Performing human-in-the-loop model validation using Labelbox and rubric-based assessments * Designing independent evaluations for individual agents and end-to-end multi-agent workflows * Exploring deterministic checks, LLM-as-a-judge evaluation, and human-review workflows Quality Engineering and Automation * Building scalable automation frameworks using Python, Pytest, Behave, and Playwright * API, UI, data-pipeline, integration, and end-to-end testing * Validation of Kafka-based streaming workflows and distributed data-processing systems * Failure simulation, schema validation, reconciliation, and data-integrity testing * Operational Acceptance Testing and production-readiness validation * CI/CD and deployment verification using GitHub Actions * Observability-driven validation using logs, metrics, and system events * IAM/RBAC boundary, entitlement, and access-control testing Data and Cloud Engineering Experience I previously worked extensively in Big Data QA across Hadoop-based ecosystems, including Spark, Hive, Kafka, Schema Registry, EMR, and Amazon S3. This experience developed my ability to analyse distributed workflows, trace failures across system boundaries, and validate large-scale data-processing behaviour. I have also tested and analysed cloud-native solutions involving AWS Lambda, EKS, IAM, DynamoDB, S3, event-driven integrations, and GCP-based document-processing workflows. Selected Engineering Contributions * Designed enterprise-scale automation frameworks using Python and Ruby * Developed structured evaluation strategies for multi-agent AI applications * Reviewed golden data and designed measurable rubric-scoring models * Created AI-assisted pull-request review plugins that generate contextual review prompts * Built engineering-insights workflows using DORA metrics * Contributed to platform plugins, integration automation, and quality-gate implementation * Led SIT, UAT, regression, and end-to-end validation for business-critical systems Technical Stack Python | Ruby | SQL | Pytest | Behave | Playwright | Google ADK | Labelbox | Promptfoo | Kafka | Schema Registry | Spark | Hive | AWS | GCP | GitHub Actions | REST APIs | PostgreSQL | GCP | I can help with: * Agentic AI and LLM evaluation strategy * Golden-dataset and test-case design * Evaluation rubric and scoring-system development * Human and automated model-evaluation workflows * Multi-agent workflow validation * Python test-automation framework development * Data-pipeline and Kafka testing * Cloud-platform and API quality engineering * Independent technical review of AI-generated deliverables If you are building an AI product, agentic workflow, data platform, or automation-heavy system, I can help you establish an evaluation approach that is measurable, repeatable, and grounded in real quality-engineering practice. Best regards, Anirban