Skip to content
Sonu / 2026

Selected work

  1. 01 · AskMyGuru

    AI Quality · Automation

    Active

    Can a system read a PRD and a Figma file and produce test cases a human would have written?

    Turning repetitive test design into a generation pipeline with human review.

    Test design per feature, including human review
    1–1.5 days→3–4 hours

    Python / Anthropic Claude SDK / Notion / Figma

    Read the case study →

  2. 02 · AskMyGuru

    AI Quality

    Active

    How do you test a system when there isn't always one correct answer?

    Measuring correctness in a system whose output is different every time.

    What the suite could detect
    Functional tests passing→Grounding failures surfaced

    Python / DeepEval / Anthropic Claude SDK

    Read the case study →

  3. 03 · Mobile Premier League

    Performance · Reliability

    How do you know a system survives a peak you can't rehearse?

    Validating capacity for traffic peaks that only happen once a year.

    Fantasy traffic before a peak event
    Unverified peak capacity→~1M RPM validated

    Locust / Kubernetes / Prometheus / Grafana / Chaos Mesh

    Read the case study →