CV

Education

• Ph.D in Computer Science, CGPA: 4.0
  Florida International University, Miami, Florida, USA, April 2013 (expected)
• M.S. in Computer Engineering CGPA: 3.8
  Western Michingan University, Kalamazoo, Michigan, USA, 2018
• B.S. in Electrical Engineering
  University of Engineering & Technology, Lahore, Punjab, Pakistan, 2014

Work experience

• Meta, Menlo Park, CA, USA
  Software Engineer Intern (Machine Learning) May 2022 - Aug 2022
  • Supervisor: Yang Xu
  • Improved Ad revenue by 5.5% by designing deep learning architectures for Ads Comments Ranking.
  • Proposed and implemented Sparse Neural Network and Transfer Learning based solutions.
  • Implemented point-wise, pairwise, and slate ranking algorithms and provided an initial evaluation
  • Increased post-click Ad conversion by 11.96% by integrating text features into the ranking platform.
  • Collaborated with multiple cross-functional teams and served as the main point of contact.
• Florida International University, Miami, FL, USA
  Graduate Research Assistant Aug 2018 - May 2023 (Expected)
  • Supervisor: Dr. Fahad Saeed
  • Created Self-Supervised Masked Language Model for Learning Protein Representation.
    • Trained the language model on a distributed cluster environment using 1 Billion protein segments.
    • Pretext protein language model using BERT improves the performance of downstream models.
  • SpeCollate: Deep Learning Network for Cross-Modal Retrieval. https://pcdslab.github.io/SpeCollate/
    • Improved protein sequencing accuracy to 95% compared to existing tools with < 70%. (Python, PyTorch)
    • Formulated SNAP-Loss: Customized quadruplet-based loss function for training SpeCollate.
    • Implemented distributed multi-GPU training of SpeCollate on SDSC Expanse supercomputer V100 GPU nodes.
    • Helped write the successful proposal for the project’s US$1 million National Institute of Health grant.
    • Awarded US$100 thousand worth of resources on XSEDE SDSC Expanse supercomputer cluster.
  • Multitask Attention Network for Predicting Peptide Properties. https://github.com/Usman095/DeepAtles
    • Designed novel deep learning attention-based architecture for learning mass spectra embeddings.
    • Reduced search space size by 90% for cross-modal retrieval using Multitask Learning. (Python, PyTorch)
  • Mentored overall 20 Undergraduate, master’s, and Ph.D. students in their research projects and career prospects
• Western Michigan University, Kalamazoo, MI, USA
  Graduate Teaching Assistant Aug 2016 - July 2018
  • Supervisor: Dr. Fahad Saeed
  • Formulated a 1.5x faster parallel network sampling algorithm using OpenCL for FPGAs. (C/C++, Big Data)
  • Created custom data structure (COPRA) for parallel removal of nodes and edges from the network.
• Xavor, Lahore, Pakistan
  Software Engineer July 2014 - July 2016
  • Supervisor: Muhammad Athar Shafiq
  • Implement HR Management Portal for Guitar Center using SharePoint. (C#, .NET)
  • Designed and implemented database merger workflow of Guitar Center and Music & Arts. (Agile Methodology)

Projects

• Winner ShellHacks 2021 Hackathon for CloudVision: A social distancing app for the visually imparied.
  • Best use of Google Cloud API award. First Prize.
  • Best Health Access and Delivery App by MITRE. Second Prize.
• TogetherAI: AI-based dating app.
  • Suggests matches based on AI-based profile images and description analysis using ResNet and GPT2.
  • Improves engagement between matches by suggesting ice-breakers based on mutual likenesses.

Skills

• Core Skills: Algorithm Design and Analysis, Deep Learning, Machine Learning, Computer Vision, Decision Trees, LSTMs, Attention Networks, Convolutional Neural Networks, Representation Learning, Two-Tower Networks (Dual Encoder Networks), Parallel Computing
• Frameworks: PyTorch, Keras, TensorFlow, SciKit-Learn, NumPy, Pandas, Jupyter Notebook
• Programming Languages: Python, C++, C, C#, Java, Cuda, OpenCL, MPI, OpenMP

Patents

Systems and methods for measuring similarity between mass spectra and peptides

Saeed, F., & Tariq, M. U. (2022). "Systems and methods for measuring similarity between mass spectra and peptides." IU.S. Patent No. 11,251,031. Washington, DC: U.S. Patent and Trademark Office..

Publications

SpeCollate: Deep cross-modal similarity network for mass spectrometry data based peptide deductions

Tariq, M. U., & Saeed, F. (2021). "SpeCollate: Deep cross-modal similarity network for mass spectrometry data based peptide deductions." PloS one. 16(10), e0259349.

Unsupervised structural classification of dissolved organic matter based on fragmentation pathways

Leyva, D., Tariq, M. U., Jaffé, R., Saeed, F., & Lima, F. F. (2022). "Unsupervised structural classification of dissolved organic matter based on fragmentation pathways.." Environmental Science & Technology. 56(2), 1458-1468.

Methods for Proteogenomics Data Analysis, Challenges, and Scalability Bottlenecks: A Survey

Tariq, M. U., Haseeb, M., Aledhari, M., Razzak, R., Parizi, R. M., & Saeed, F. (2020). "Methods for proteogenomics data analysis, challenges, and scalability bottlenecks: a survey." IEEE Access. 9, 5497-5516

Power-efficient and highly scalable parallel graph sampling using fpgas

Tariq, U., Cheema, U. I., & Saeed, F. (2017, December). "Power-efficient and highly scalable parallel graph sampling using fpgas." International Conference on ReConFigurable Computing and FPGAs (ReConFig) (pp. 1-6). IEEE.

Graph Theoretic Approach for the Analysis of Comprehensive Mass-Spectrometry (MS/MS) Data of Dissolved Organic Matter

Molecular level characterization of DOM along a freshwater-to-estuarine coastal gradient in the Florida Everglades

Awards and Talks

• Winner ShellHacks 2021 Hackathon for CloudVision. https://devpost.com/software/cloudvision
• Winner 3MT Poster Presentation on SpeCollate at SCIS FIU Research Day Event 2019.
• Awarded Rising Star of the Year 2015 as a Software Engineer at Xavor Corporation
• Guest Speaker “Supercomputing and Machine Learning for Big Biomedical Data” at the Department of Electrical Engineering, University of Engineering and Technology, Lahore, Pakistan. 2021 https://www.youtube.com/watch?v=dUvaxjDeBmM
• Presented “Parallel Sampling-Pipeline for Indefinite Stream of Heterogeneous Graphs using OpenCL for FPGAs.” at 2018 International Conference on Big Data.
• Presented “Power-efficient and highly scalable parallel graph sampling using FPGAs.” at 2017 International Conference on ReConFigureable Computing and FPGAs.
• Presented “Predictive activation for localization using minimal data-fusion in MANETs.” at 2013 International Conference on Information and Communication Technologies.

Leadership and Extra-Curricular Activities

• Tech Lead, AI/ML Community with more than 230 active members as part of Google Developer Student Club.
• Leader of Deep Learning paper reading club with 10 active researchers.
• Reviewer for IEEE Access, PLOS, and Distributed Ledger Technologies.
• Volunteer coordinator at Midscale Experimental Research Infrastructure Forum (MERIF 2020).
• Volunteered organizer at International Multi-Topic Conference 2013.
• Member Upsilon Pi Epsilon (UPE), UPE Leap, UPE SparkDev, and UPE Code student organizations.
• Designed IEEE UET Web Portal as a member of the IEEE Student Chapter during my bachelor’s degree.
• Organized several events and workshops (Programming Competitions, LaTeX, MATLAB Workshops) as a member of the IEEE Student Chapter during my bachelor’s degree.