Logo

Yawar Ashraf

I'm a fourth Year student at the Unversity of Toronto studying Engineering Science in the Machine Intelligence option with a Minor in Robotics. I have extensive experience in AI, software and robotics through many projects and internships.

Experience

Computer Integrated Manufacturing Lab

Research Assistant

May 2023 - April 2024
  • Working on a decentralized market-based task allocation algorithm to optimize trajectory planning for heterogeneous Multi-Robot Systems while leveraging a dynamic probabilistic lost person's model at Professor Beno Benhabib's Computer Integrated Manufacturing Lab (CIMLab).

AMD

Memory Design and Validation Intern

May 2022 - April 2023
  • Setup and executed stressful memory tests and performed careful GDDR6 chip tuning from multiple memory vendors based on a deep analysis of characterization results.
  • Designed and Developed multiple memory, bandwidth and power measurement analysis tools that reduced processing time of key tasks by 40%-symbol and saved over 80+ hours every month.
  • Conducted an internal study on GDDR6 memory devices using an oscilloscope, performing Spread Spectrum and Jitter analysis, which informed design considerations for GDDR7.
  • Researched and developed a ML pipeline for predicting passing phase and voltage regions for memory characterization.

Typebrite

Co-Founder and Chief Technical Officer

February 2023 - Present
  • Led multiple client engagements, successfully secured an enterprise client contract worth over 35,000 CAD for API documentation & developer environment solutions tailored to enhance developer workflows & client engagement.
  • Steered Typebrite to specialize in transforming raw APIs into engaging documentation using Large Language Models and git styled developer workflows and collaborative documentation generation.

Networks Lab

Research Assisstant

May 2021 - August 2021
  • Received a 2021 NSERC award for Summer Research at Prof Jorg Liebeherr's Networks Lab.
  • Designed, built and tested a power-saving circuit that was designed to allow for 6-12 months of Long Range communication with no more than 2 AA batteries.
  • Contributed to the development and testing of a Multi-Hop self organizing LoRa 32u4 network for surveying and collecting waterfall data across a village in Mexico to help design and install rain water collection systems.
  • Designed and built outdoor electronic circuit enclosures in compliance with IP65 waterproof rating. Prepared detailed enclosure and circuit design documents for replicability and to be implemented in Mexico.

Autoronto

Simulation and Deep Learning Accelaration Team Member

January 2021 - September 2022
  • Developed a noise modelling technique for the car’s camera by building a custom Generative Adversarial Network (GAN) to accurately map from noisy images to clean images via a Denoising Convolutional Neural Network (DnCNN)
  • Implemented a data serialization technique to leverage hardware Accelarated Inference.
  • Maintained testing track map in OpenStreetMap using JOSM and was one of the leaders for leading to generate and label custom Object Detection datasets
  • 1st Place Winner of 2022 SAE Autodrive Challenge II

Content Turbine

Software Engineer

March 2021 - September 2021
  • Synthesized an ETL (Extract, Transform, Load) framework across several database platforms such as MySQL, PostgreSQL, Snowflake and BigQuery.
  • Implemented GraphQL, reducing frontend data over-fetching and improving application performance by enabling precise data retrieval, leading to a more responsive user experience.
  • Contributed to database management tasks, including schema design, optimization, and maintenance, ensuring data integrity and efficient querying for analytics purposes.

Town of Oakville

Squash Instructor

February 2023 - Present
  • Designed and implemented structured training programs for group sessions, incorporating skill progression, fitness components, and strategic gameplay.
  • Successfully managed groups with varying skill levels, implementing differentiated instruction to address individual needs, enabling a rewarding learning experience for players of all levels.

McDonalds

Team Member

September 2018 - August 2020
  • Executed various kitchen tasks, including food preparation, cooking, and assembly, ensuring timely and accurate order fulfillment while maintaining high standards of food safety.

PIA

Squash Colt

July 2014 - June 2018
  • Achieved notable success by reading the #8 rank in Pakistan in the BU15 category as a competitive squash player.

Projects

AMINet - Blood Clot Detection using Vision Transformers

December 2023
  • Used cutting edge vision transformers (e.g. EfficientFormerV2, MobilNetV2) to build a attention-leveraged deep learning architecture to detect brain regions affected by a stroke given medical imaging (mCTA) scans.
  • Achieved commercially competitive results while working with Andromeda Medical Imaging (AMI).
  • Acheived less than 10s inference time while using a single CPU (less 8GB RAM) and outperformed previous solution to achieve an 87% F1 score.
  • View Project

Recycle Right

December 2021
  • This project aimed to tackle the problem of < 5% plastic recycling in Ghana by focusing on the waste collection system.
  • Designed and prototyped a bin overflowing mechanism and an app with a cloud implementation to keep track of overflowing bins in a neighborhood.
  • implemented our prototype using an Arduino Uno, BN-180 GPS Module, RCWL-1601 Ultrasonic sensor, HC-05 Bluetooth module, LoRa (Long Range) transciever, Firebase backend, and React frontend
  • View Project

Stock Market Prediction via Twitter Sentiment Analysis

April 2022
  • Engineered a custom AI model using the LSTM architecture and BPTT for S&P value forecasting leveraging key market indicators and sentiment analysis performed on tweets.
  • The data scraper was adapted using Twint and the sentiment analysis post Tokenization and Lemmatization was performed using Vader.
  • View Project

Invasive Ductal Carcinoma Prediction Using Deep CNNs

April 2021
  • Orchestrated the preprocessing of a massive breast histopathology dataset, comprising over 550,000 50x50 pixel patches, by organizing, redistributing and cleaninig data to reduce bias and skew.
  • Engineered two advanced IDC detection models: a stride-based end-to-end Deep CNN Model and a pre-trained AlexNet architecture with tailored input adaptation achieving an accuracy of 86.4%.
  • View Project

Self Driving Autonomous Robot

December 2023
  • Designed and impleneted an SAE Level 4 autonomous rover to navigate a known maze and perform a block delivery and block pick up from an initially unkown and random position.
  • Leveraged both Histogram Localization and Particle Filter algorithms to perform robot localization and consequential path planning.
  • Performed multiple design lifecycles from ideation to CAD and to real world testing in a 3-tiered milestone based design approach to build both the robot structure, electrical components, block delivery system and required software in a seamless manner.
  • View Project

Hierarchical Self Driving Car Motion Planner: CARLA

August 2020
  • Used finite state machines to select safe behaviors to execute, and design optimal, smooth paths and velocity profiles to navigate safely around obstacles while obeying traffic laws and implemented in the CARLA Simulator.
  • Built occupancy grid maps of static and dyanmic elements in the environment and use them for efficient collision checking.
  • Performed continuous curve path optimization as a two point boundary value problem to minimize deviation from the desired path while satisfying curvature constraints.
  • View Project , View Certificate