Muhammad Qasim

• Natural Language Processing Engineer - Quarrio

  September 2016 – Present

  Working on the development of Deterministic Semantic AI utilizing multi-level parsers and grammars that understand language nuances, from general English to industry-specific jargon. AI system enables anyone, regardless of technical background, to interact with data seamlessly. Unlike probabilistic AI models that guess the best possible answers, deterministic approach ensures precise, contextually accurate responses every time.

  Natural Language Processing

  AI system understands questions asked on structured data in plain English language. Question as understood by the system is provided to the user in plain English as well, making validation easy.

  Cognitive Language Processing

  AI system navigates through an abstraction layer that identifies and accesses relevant data sources. It generates precise query statements, retrieves information, and presents it in an easy-to-understand format.

  Contextual Understanding

  AI system understands context of a question. For a given question, it establishes references with previously asked questions to provide response using the context in which that question was asked. AI system engages with user to get additional information before answering a question in case it cannot completely interpret the question.

• Research Associate - Center for Language Engineering, University of Engineering & Technology, Lahore

  December 2014 – September 2016

  Worked on development of speech corpora, Automatic Speech Recognition (ASR) systems, and spoken dialog systems for the Urdu language.

  Automatic Speech Recognition (ASR) systems

  Worked on development of ASRs for names of cities, numbers, days of week, etc. for the Urdu language.

  Spoken dialog systems

  Worked on development of spoken dialog systems for weather information and bus reservation systems using speech interface. These systems were deployed over telephone helplines where users could call to get the required information. Weather information system was made the national weather helpline of Pakistan Meteorological Department where it was mainly used by the farmers. Several Human-computer interaction (HCI) interventions (multi-modal input, user adaptation and context-specific help) were made to improve the performance of the system. Based on test period spanning 23,997 phone calls from 6893 users over 10 months, the task success rate increased to 96.3\%.

  Speech corpora

  Worked on designing, recording, and pre-processing speech corpora for the Urdu language for different domains.

• Teaching Assistant - Computer Engineering Department, University of Engineering & Technology, Lahore

  September 2017 – December 2017

  Prepared and graded lab manuals and weekly assessments. Conducted weekly tutorial session to help students with the learning material and assignments. Supervised term projects of students.

  Undergraduate-level Digital Signal Processing course

  Graduate-level Speech Processing course

• Lecturer - Punjab Higher Education Department, Lahore

  July 2021 – December 2021

  Taught high school level Computer Science courses.
  

Education

• MS Computer Engineering - University of Engineering & Technology Lahore, Pakistan
  CGPA: 3.92/4.00
  Thesis: Speech Emotion Recognition for Urdu Language
  
• B.Sc. Electrical Engineering - University of Engineering & Technology Lahore, Pakistan
  CGPA: 3.07/4.00
  Final Year Project: Speech Controlled Vehicle
  

Research Area

Artificial Intelligence, Machine Learning, Natural Language Processing, Speech Recognition, Speech Emotion Recognition.

• Learn More