Skip to content
View mwasifanwar's full-sized avatar
🎯
Focusing
🎯
Focusing
23 followers · 0 following

Achievements

Achievement: StarstruckAchievement: QuickdrawAchievement: Pull Shark

Achievements

Achievement: StarstruckAchievement: QuickdrawAchievement: Pull Shark

Block or report mwasifanwar

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
mwasifanwar/README.md

Personnel file of Muhammad Wasif Anwar. Your AI passed the demo. These are the tests it takes next.

Muhammad Wasif Anwar

AI engineer in Hannover, Germany. I build LLM applications, AI agents, voice AI and automation, and I care most about the part that comes after the demo: real users, messy data, and nobody watching.

Every AI I build gets a probation period. This profile is where the tests for that probation become open source.

The AI Probation toolkit

Nine questions every AI agent should answer before it talks to customers alone, each turned into a small tool you can run on your own agent. One repo per question, plus one command that runs them all.

Question Tool Status
§01 Can it change without anyone touching it? driftwatch planned
§02 Can it say no, four times in a row? pushback building now
§03 Are its permissions based on reversibility? undo-or-ask planned
§04 Can a stranger rewrite its instructions? canary-docs planned
§05 Was it tested on your customers, or on your team? ticket-replay planned
§06 What happens when two documents disagree? kb-conflicts planned
§07 Does the customer have to start over with a human? handover planned
§08 Can you explain one answer from last Tuesday? why-ledger planned
§09 Are you measuring resolution, or just silence? silence-audit planned
✱ All nine, one scorecard ai-probation planned

Every tool ships with a real example run, tests, and a section called Known limitations. If a README claims something, the repo can show it.

Want to run the review by hand first? It takes five minutes: mwasif.dev/review

How I work

  1. I'll ask what should happen when the AI is wrong. Early. More than once.
  2. I'd rather show you a failure in week one than hide it until month three.
  3. "Small" is a hypothesis, not a measurement.
  4. Every decision gets its reason written down.
  5. If it can't be undone, a human approves it.

Tools I work with

  • AI and LLMs: LangChain, RAG, AI agents, model evaluation, PyTorch, TensorFlow
  • Voice and vision: speech-to-text, text-to-speech, YOLOv8, OpenCV, OCR
  • Backend and data: Python, FastAPI, Node.js, Express, PostgreSQL, MongoDB
  • Frontend and delivery: TypeScript, React, Docker, CI/CD, AWS

On file elsewhere

One of my AI agents once filed a formal complaint about me. HR took no action.
Read the thread

End of file.

Pinned Loading

  1. IndustrialPrognosisAI IndustrialPrognosisAI Public

    Advanced Condition Monitoring and Remaining Useful Life Prediction Framework using Deep Learning for Industrial Equipment Prognosis and Predictive Maintenance

    Python 2

  2. llm-mastery llm-mastery Public

    The most comprehensive educational resource on Large Language Models ever created. A guide systematically building understanding from absolute fundamentals to cutting-edge research.

    2

  3. AutoCV AutoCV Public

    One-click computer vision - automatically trains models on your images without coding.

    Python 1

  4. ChurnVision-AI ChurnVision-AI Public

    AI-powered churn prediction pipeline integrating clustering, association rule mining, and ensemble learning,built for interpretable customer retention analytics.

    Jupyter Notebook

  5. DeepForge DeepForge Public

    Advanced multi-model deepfake detection framework combining ensemble machine learning with deep neural networks for robust image authentication and media forensics.

    Python

  6. NeuroScan-AI NeuroScan-AI Public

    NeuroScan-AI is an advanced document-understanding engine built with modern computer vision and OCR pipelines. It performs smart perspective correction, illumination normalization, and adaptive enh…

    Python 1