Ujjwal Kumar Singh beinghumantester

Still figuring out where you land? I'm

Quality Engineer
Breaking software before users have to.

Agentic AI Builder
Teaching AI agents to investigate, not just execute.

Community Builder
Writing and speaking on QA and applied AI.

Issue Tracker — About Me

#	Title	Labels
#001	Doesn't fully trust "it works on my machine"	`skepticism` `evidence-based`
#002	Prompts AI agents to investigate failures, not just run scripts	`agentic-testing` `ai`
#003	Building an AI interview-prep assistant that grades reasoning	`side-project` `interviews`
#004	Writes about QA + AI on Substack instead of keeping notes private	`writing` `substack`
#005	Used to just write test cases. Now asks why things break	`career-growth`

Journey So Far

2019 ─ Started as a manual tester
        │ Learned to break things before users could
        │
2021 ─ Moved into automation
        │ Selenium, Pytest, CI pipelines
        │
2023 ─ Started asking "why," not just "what failed"
        │ Root-cause diagnostics over pass/fail reports
        │
2024 ─ Started writing prompts, not just test cases
        │ Prompt engineering for structured, testable AI output
        │
2025 ─ Evaluating LLMs, not just web applications
        │ Testing model behaviour, reasoning, and failure modes
        │
2026 ─ Now: teaching AI agents to investigate
        └ Agentic testing, autonomous root-cause analysis

How My AI Agent Investigates a Failure

flowchart LR
    A(["Signal detected<br/>failure / anomaly / crash"]) --> B["Agent ingests context:<br/>logs · stack trace · DOM snapshot"]
    B --> C["Form hypothesis on root cause"]
    C --> D["Reproduce in sandbox"]
    D --> E{"Hypothesis confirmed?"}
    E -- No --> F["Discard & generate<br/>new hypothesis"]
    F --> C
    E -- Yes --> G["Classify root cause<br/>flaky · regression · env · code"]
    G --> H{"Confidence high enough?"}
    H -- No --> I["Flag for human review"]
    I --> J(["Engineer investigates"])
    H -- Yes --> K["Auto-generate report:<br/>cause · evidence · suggested fix"]
    K --> L(["Handed off to dev queue"])

    style A fill:#1B6E92,color:#fff
    style F fill:#0F6E56,color:#fff
    style I fill:#D85A30,color:#fff
    style K fill:#2E9E7A,color:#fff
    style L fill:#2E9E7A,color:#fff

_{The loop behind my Agentic Testing Framework project — an AI agent doing the root-cause investigation on its own, escalating to a human only when it isn't confident.}

Test Coverage Report — Skills

$ pytest --cov=beinghumantester --cov-report=term

Automation & Testing
  Python           ████████████████████  96%
  Selenium         ███████████████████░  92%
  Playwright       █████████████████░░░  85%
  Pytest           ██████████████████░░  88%

AI & Backend
  FastAPI          ██████████████░░░░░░  70%
  PyTorch          ████████████░░░░░░░░  60%
  OpenAI / LLMs     ████████████████░░░░  80%
  MCP              █████████████░░░░░░░  65%

DevOps & CI/CD
  Docker           ███████████████░░░░░  75%
  Jenkins          █████████████░░░░░░░  65%
  GitHub Actions   ████████████████░░░░  80%
  Git              ████████████████████  98%

------------------------------------------------
TOTAL                                     80%   PASSED

Active Experiments & Flagship Projects

Pytest + Selenium Diagnostics

Automated failure root-cause analysis

Captures DOM state, browser logs, and network telemetry automatically on failure
Turns raw stack traces into actionable reports

AI Interview Prep Assistant

Evaluating reasoning over memorization

Scenario-driven AI interviewer
Scores problem-solving pathways, not keyword matches

Agentic Testing Framework

Autonomous QA agents

Runs exploratory test loops on its own
Classifies crashes and reasons about root cause

Field Notes & Media

Thought leadership in QA & AI

Long-form essays on automation architecture
Talks & workshops on AI in testing

GitHub Telemetry

Words I Test By

"Information is power. But like all power, there are those who want to keep it for themselves." — Aaron Swartz

$ ./run_tests.sh
> When I don't test the code,
> I code to test the code.

Let's Connect & Collaborate

I'm always open to discussing test architecture, agentic AI frameworks, or quality engineering strategy.

Navigation Menu