Urwa Muaz, Computer Vision Developer in Lahore, Punjab, Pakistan
Urwa Muaz

Computer Vision Developer in Lahore, Punjab, Pakistan

Member since October 1, 2019
Urwa is a Fulbright Scholar and Data Science graduate from New York University. He loves to leverage machine learning to solve practical problems and enjoys challenging, research-oriented positions. He has taken part in several machine learning projects, with deep learning-based Computer Vision being his primary expertise.
Urwa is now available for hire




Lahore, Punjab, Pakistan



Preferred Environment

GitHub, PyTorch, Jupyter, Linux

The most amazing...

...project I've designed and implemented was a novel loss function to reduce gender bias in a neural language model, achieving state-of-the-art results.


  • Machine Learning Consultant

    2019 - PRESENT
    Urban Complexity Lab, New York University
    • Created low dimensional representation from high dimensional transport data using autoencoders.
    • Worked on a non-linear traffic prediction model for the outflow of traffic from JFK.
    • Worked on anomaly detection pipelines based on Gaussian mixture models.
    Technologies: Scikit-learn, PyTorch, Python
  • Senior Machine Learning Engineer

    2019 - PRESENT
    • Worked on vehicle detection and tracked research and development for an automatic traffic violation detection system based on road facing cameras.
    • Re-implemented a state-of-the-art solution for vehicle tracking by detection proposed in the paper "Extending IOU Based Multi-Object Tracking by Visual Information."
    Technologies: OpenCV, PyTorch, Python
  • Computer Vision Intern

    2018 - 2019
    Applied Research in Government Operations (ARGO)
    • Wrote scripts to collect image datasets from open source web resources like OpenStreetMap (OSM) and Google street view.
    • Utilized Microsoft custom vision API for data annotation and model training.
    • Developed a model used to build a report to analyze the comparative distribution of taxis on roads across Manhattan.
    Technologies: Python, Microsoft Visio
  • Computer Vision Engineer

    2016 - 2018
    • Led a team to develop a Computer Vision product that uses the security infrastructure (cameras) in retail stores to provide customer behavior analytics such as footfall, dwell time in different zones, and heat maps.
    • Integrated a repeat customer identification system based on FaceNet Siamese embeddings. Improved the accuracy from 60% to 90% by writing a custom temporal tracking layer over the base single shot face detector.
    • Worked on a .NET web application for retail store performance management, which involved near real-time integration of heterogeneous data sources and extensive dashboarding.
    • Involved in multiple proof of concept solutions, including retail demand forecasting and a product-level price recommendation engine.
    Technologies: .NET, Python
  • Research Assistant

    2015 - 2017
    Biomedical Informatics Research Lab, LUMS
    • Developed a next-generation top-down protein search engine.
    • Created parallel versions of several algorithms developing respective CUDA kernels for execution on Nvidia GPUs.
    Technologies: CUDA, .NET


  • Reducing Gender Bias in Word-level Language Models with Gender-Equalizing Loss Function

    This paper proposes a novel method to address gender bias in the neural language models. We introduce a penalty term to the objective function of the language model to penalize the discrimination against gender. This method benefits from being simple, intuitive and can be easily incorporated into any text generation model. The proposed model’s performance was evaluated using multiple fairness metrics as well as perplexity showing that this method, when trained with counterfactual data augmentation, outperforms all existing techniques in the literature.

  • Semi-supervised Image Classification with Low Resource Labelled Data

    I attended Yann Lecun's Deep Learning class at NYU last year, and this was my term project from it.

  • Academic Excellence Award 2019 NYU CUSP and Valedictorian Speech

    I was awarded the academic award for the class of 2019 at NYU CUSP. This also gave me the honor to address the esteemed audience at the convocation ceremony as a valedictorian speaker.

  • Fare Evasion Detection using Computer Vision

    I proposed and presented a cost-effective automated system to measure (rather than estimate) the fare evasion in New York subways using security camera footage to MTA.

  • Bronze Medal in 42nd International Physics Olympiad

    This was a global high school competition where about 90 countries participated.

  • Human Steering Behaviors Study using Computer Vision

    We wanted to observe if the steering behaviors described are observed in urban settings. We collected video data from a few urban spaces around metro tech to observe the pedestrian mobility tracks in those spaces and used Computer Vision to perform the analysis.

  • Spatial Geography of Emotions in New York

    We collected geo-tagged data from Flickr and Twitter and performed textual and visual sentiment analysis of different locations in New York.

  • Data Driven Behavioral Analysis of Smokers

    We used health survey data to identify socioeconomic health and lifestyle conditions that influence smoking behavior. Additionally we investigated to what extent can smoking behavior be predicted on individual level based on these factors. We had some interesting findings which are listed in the provided link.

  • NLP tool for Automated Discovery of Datasets, Research Fields, and Methods from the Raw Text of Publications

    ● Named entity recognition and a convolutional neural network
    classifier pipeline was used to identify the sentences that mention datasets.
    ● Ngram and cosine similarity based pipeline to match mentions to known datasets.
    ● Document similarity approaches were used for research field identification.

  • Semi-supervised Image Classification With Unlabeled Data (Publication)
    Supervised learning is the key to computer vision and deep learning. However, what happens when you don’t have access to large, human-labeled datasets? In this article, Toptal Computer Vision Developer Urwa Muaz demonstrates the potential of semi-supervised image classification using unlabeled datasets.


  • Other

    Computer Vision, Machine Learning, Deep Learning, Natural Language Processing (NLP)
  • Languages

    Python, SQL
  • Libraries/APIs

    TensorFlow, PyTorch, Keras, OpenCV, Scikit-learn
  • Platforms

    Linux, CUDA
  • Frameworks

  • Tools

    Jupyter, GitHub, Microsoft Visio
  • Paradigms

    Agile Software Development
  • Storage



  • Master's degree in Data Science
    2018 - 2019
    New York University - New York, NY, USA
  • Bachelor's degree in Elctrical Engineering and Computer Science
    2012 - 2016
    University of Engineering and Technology - Lahore, Pakistan


  • Deep Learning Specialization
    JUNE 2018 - PRESENT
  • Data Science Professional Project
    MARCH 2017 - PRESENT

To view more profiles

Join Toptal
Share it with others