Nayef Ahmed

ENGINEER  |  RESEARCHER  |  DEVELOPER

About

HELLO WORLD

Hey, I'm Nayef! I'm passionate about software development, artificial intelligence, and autonomous vehicles and love working on projects where these fields intersect. I'm also a personable team player with an aptitude for technology, an intuition for solving challenging problems, and a strong track record in people leadership.

Experience

PREVIOUS WORK

SOFTWARE ENGINEER

META AI
Python, C++, Pytorch, BigQuery  |  New York, NY  |  Jun 2021 - Present
  • Tech leading a team of 4 engineers to deliver a multi-modal representation model capable of understanding images, videos, text, and audio by jointly training a vision/text encoder along with a transformer-based fusion layer
  • Developed new components (datasets, transforms, models) as a core maintainer of the PyTorch Text library, engaged in feature discussions and solicited feedback from both the open source and internal communities
  • Managed a summer intern to develop a transformer-based encoder-decoder model (T5) for text generation tasks; wrote a detailed project plan, provided mentorship, pushed for customer adoption, and represented the intern during final evaluations
  • Increased library adoption 300% YOY by driving quarterly PyTorch releases and broadcasting new features to researchers

SOFTWARE ENGINEER INTERN

FACEBOOK AI
Python, C++, Pytorch, TorchScript  |  New York, NY (Remote)  |  May - Sep 2020
  • Worked with the AI Platform team to design and implement high-performance NLP building blocks (Tokenizer, Vector, Vocab) for PyTorch
  • Profiled CPU performance of NLP preprocessors on several datasets using Flame Graphs and applying concurrency to decrease load times by over 60x
  • Won 2 awards for creating a Workplace VC “quickchat” feature at the Global Summer Hackathon, receiving overwhelming interest to ship to production
  • Supported upcoming Pytorch releases to facilitate NLP research and demonstrate the feasibility of new features for internal adoption

SOFTWARE ENGINEER INTERN

LYFT LEVEL 5 (AUTONOMOUS VEHICLES)
Python, C++, TensorFlow, OpenCV  |  Palo Alto, CA  |  Jan - Apr 2020
  • Trained new classifier with the Perception team capable of detecting traffic light bulb shapes with 90%+ accuracy on an extremely unbalanced dataset
  • Designed a regression test suite capable of providing a binary decision to engineers when deciding whether to release a new model to the autonomous vehicle flee
  • Created a test creation GUI that gives engineers finer control when selecting relevant obstacles from an image which cannot be regressed on in the future

SOFTWARE ENGINEER INTERN

FACEBOOK AI RESEARCH
Python, C++, OpenCV, PyTorch  |  Menlo Park, CA  |  May - Aug 2019
  • Worked within the Applied Research Computer Vision team to create a face detection and tracking pipeline running on all Facebook videos in real-time
  • Benchmarked precision and recall of various detection models (Mask R-CNN, DFSD, Viola-Jones) and tracking algorithms on a 600k SOA video dataset
  • Closely collaborated with leading researchers (Ross Girschick and Yufei Wang) to anonymize faces in the video dataset using detection and tracking results

SOFTWARE ENGINEER INTERN

MICROSOFT CORPORATION
C#, Redis, Azure  |  Redmond, WA  |  Aug - Dec 2018
  • Worked within Azure DevOps team to create an Ownership Enforcer (OE) service that enforces PR code reviews for compliance and security assurance
  • Improved performance of OE service by designing multi-layered Redis Cache pipelines from the ground up thus reducing VSTS API calls by 70%
  • Redesigned existing cache systems to optimize lookup time by ensuring scalability across multiple shards to support future expansion of OE service

PATH PLANNING ENGINEER

WATONOMOUS DESIGN TEAM (SELF-DRIVING CAR)
OpenCV, ROS, Rviz, LiDAR, Radar  |  Waterloo, ON  |  Sep 2017 - Aug 2018
  • Worked within Sensor Fusion subteam to transform a Chevrolet Bolt EV into a level 4 autonomous vehicle for the SAE AutoDrive Challenge
  • Implemented path visualization algorithms using ROS and Rviz to display the path of the car generated using a trajectory rollout approach
  • Developed core algorithms for cost map generation using a 2-layered occupancy grid system for dynamic replanning

SOFTWARE ENGINEER INTERN

MICROSOFT CORPORATION
C#, C++, C++/CX, XAML, Unity, HoloLens  |  Vancouver, BC  |  Jan - Apr 2018
  • Worked on Mixed Reality Viewer team to create a default Windows and HoloLens app for viewing 3D content with over 3 million monthly users
  • Integrated 3D asset conversion and optimization pipeline to support all 3D file formats and ensured it adhered to the MVVM design pattern
  • Collaborated with UI/UX designers and data analysts to implement a content promotion system that promotes key app features to new users

FULL-STACK DEVELOPER

UNICELL BODY COMPANY
Java, MySQL, Laravel, Vue.js, React Native  |  Toronto, ON  |  May - Aug 2016
  • Expanded company’s business-facing website using LAMP stack, Laravel, and Vue.js used by over 1000 employees
  • Automated core data processing using Selenium and Java, resulting in 60% increased efficiency in chassis inventory management
  • Iterated on internal quoting tools using Chart.js to improve ease of use and allow for better visualization of company performance

Projects

PERSONAL ENDEAVORS

WRAP

WHITEBOARD RECOGNITION AND ANALYSIS PLATFORM
Python, C++, GCP, Firebase, React Native
  • Created a cross-platform mobile app to record and upload lectures to the cloud
  • Processed lectures in real-time and applied heuristics, and computer vision to digitize video into formatted lecture notes

FLAPPY GOOSE

RTOS GAME IN AN EMBEDDED ENVIRONMENT
C, Keil RTX RTOS, LPC1768 Microcontroller
  • Recreated the Flappy Bird game using Keil MCB1700 board in a real-time OS
  • Used task synchronization and interrupts to update LCD graphics, player score, game physics, and handle peripheral I/O simultaneously

STUDY SPACE

PENNAPPS (GOOGLE PRIZE WINNER)
Android Things, Java, Firebase, NXP Microcontroller
  • Created IoT device along with companion app to display the number of people in specific locations on campus
  • Determined the occupancy of areas using ‘Android Nearby’ to discover number of wireless devices within range

SMART GLOVE

IEEE HARDWARE HACKATHON  |  3RD PLACE WINNER
Arduino, C, Gyro/Accselerometer
  • Designed glove to wirelessly control IoT devices using simple gestures
  • Implemented gesture recognition by combining data from capacitive force sensors and 3-axis gyroscope/accelerometers
  • Controlled devices such as lights and speakers by sending commands through TCP local WiFi socket using two Arduino Nanos

DOG WHEELCHAIR LIFT

ROBOTICS PROJECT
C++, RobotC, AutoCAD
  • Constructed wheelchair lift to allow handicapped dogs to scale obstacles with heights up to 20cm
  • Implemented automatic deployment/stoppage of lift sequences by detecting obstacles using gyroscope and ultrasonic sensors