Skip to content
View alinawazcode's full-sized avatar
🎯
Focusing
🎯
Focusing
5 followers · 4 following

Achievements

Achievement: Pull Sharkx2Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Pull Sharkx2Achievement: YOLOAchievement: Quickdraw

Block or report alinawazcode

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. 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
alinawazcode/README.md
Ali Nawaz — Python Developer, ML and NLP Engineer. Open to opportunities.
Typing animation: Python Developer, ML and NLP Engineer; model training, evaluation, APIs; from notebook to production
LinkedIn GitHub Hugging Face PyPI

Open to opportunities Stack: Python, PyTorch, FastAPI Pakistan Profile views

01 About

# profile.yaml
name: Ali Nawaz
role: Python Developer · ML & NLP Engineer
status: open to opportunities
strengths:
  - fine-tuning and evaluating transformer models
  - shipping models as SDKs, REST APIs and demos
  - secure, tested backends (Django, FastAPI)
how_i_work: [reproducible runs, baselines first, honest reporting]
shipped: [PyPI package, Hugging Face models and Spaces, Dockerized apps, CI pipelines]

I'm a Python developer who takes machine learning work from a research notebook to something other people can install, call and trust. I build the whole chain: labeled datasets, training and evaluation, then packaging, APIs and deployment. My NLP projects focus on low-resource languages, which taught me to benchmark carefully and report negative results when the numbers demand it.

02 What I Bring

🧠 Machine Learning & NLP

Fine-tuned XLM-RoBERTa and classical TF-IDF + LinearSVC models, trained tokenizers, and benchmarked against 14 baselines. Seeded, logged runs so results can be reproduced.

⚙️ Backend & APIs

FastAPI and Django services, including a zero-knowledge document platform with AES-256-GCM encryption, single-use QR passes and rate limiting.

🚀 Delivery & Quality

Docker, GitHub Actions CI, automated tests, a PyPI package, Hugging Face releases and documentation sites, so work ships and stays maintainable.

03 Toolchain

Tech stack: Python, JavaScript, HTML5, CSS3; PyTorch, Transformers, scikit-learn, SentencePiece, FAISS, CLIP; Django, FastAPI, Gradio, React, Vite, Tkinter; Git, GitHub Actions, Docker, HF Spaces, PyPI, PythonAnywhere, Railway

04 Featured Projects

Sindhi NLP

Transformer sentiment analysis with SDK and API

Sentiment analysis (Sindhi) with a fine-tuned XLM-RoBERTa model, a Python SDK, and a FastAPI service. Every run is seeded and logs its config, metrics, and data hashes.

Python PyTorch XLM-RoBERTa FastAPI Docker

Repository Model PyPI

Sindhi Sentiment Dataset

Classical ML classifier and labeled dataset

Dual TF-IDF + LinearSVC classifier trained on a fully human-labeled 4,420-sentence dataset: 91.7% accuracy and 0.918 macro F1, with a transparent writeup of why accuracy dropped from v1.

Python scikit-learn TF-IDF LinearSVC

Repository Live Demo

Passli

Zero-knowledge document sharing

High-security document sharing with AES-256-GCM encryption, single-use QR passes, rate-limiting defense, granular expiration controls, and zero server-side plaintext exposure.

Python Django AES-256-GCM Docker GitHub Actions

Repository Docs

Sindhi Tokenizer

Corpus-driven subword tokenizer

SentencePiece Unigram tokenizer trained on 1.46M+ sentences and benchmarked against 14 baseline tokenizers, with a 0.000% normalized-length loss round-trip check.

Python SentencePiece Gradio GitHub Actions

Repository Model Docs

Explore all repositoriesView on GitHub

05 GitHub Stats

GitHub stats: stars, commits, pull requests, issues, public repositories and followers Most used languages across public repositories

Contribution snake animation Contribution snake, generated every 12 hours by GitHub Actions.

06 Recent Activity

Latest commits, pull requests and releases
  1. 🚀 Published release Sindhi Tokenizer v1.0.0 — First Stable Release in alinawazcode/sindhi-tokenizer
  2. 🎉 Merged PR #15 in alinawazcode/sindhi-nlp
  3. 💪 Opened PR #15 in alinawazcode/sindhi-nlp
  4. 🎉 Merged PR #14 in alinawazcode/sindhi-nlp
  5. 💪 Opened PR #14 in alinawazcode/sindhi-nlp
  6. ❌ Closed PR #153 in abhiksark/pythonlings
  7. 🎉 Merged PR #24 in alinawazcode/passli
  8. 💪 Opened PR #24 in alinawazcode/passli
  9. 🎉 Merged PR #23 in alinawazcode/passli
  10. 💪 Opened PR #23 in alinawazcode/passli

Updated automatically every 6 hours by GitHub Actions.

07 Connect

Looking for a Python / ML engineer? Let's talk.
The fastest way to reach me is LinkedIn.

LinkedIn GitHub Hugging Face PyPI Zenodo
Built with Markdown, SVG, and GitHub Actions

Pinned Loading

  1. sindhi-nlp sindhi-nlp Public

    Reproducible Sindhi sentiment analysis with XLM-RoBERTa, a Python SDK, and a FastAPI service.

    Python 2

  2. Sindhi_Sentiment_dataset Sindhi_Sentiment_dataset Public

    Sentiment classifier for Sindhi text (dual TF-IDF + LinearSVC) — v2: 91.7% acc / 0.918 macro F1 on a fully human-labeled 4,420-sentence dataset. Includes a transparent writeup of why accuracy dropp…

    Python 5

  3. passli passli Public

    High-security zero-knowledge document sharing platform with AES-256-GCM encryption, single-use QR passes, rate-limiting defense, granular expiration controls, and zero server-side plaintext exposure.

    Python 2 1

  4. sindhi-tokenizer sindhi-tokenizer Public

    production-grade SentencePiece Unigram subword tokenizer for Sindhi (Perso-Arabic script) with 100% character coverage, byte-fallback, and 0.0000% character loss. Rigorously benchmarked across 17 m…

    Jupyter Notebook