Armaghan Sarvar

Experience

• 

  Amirkabir University of Technology - Tehran Polytechnic

  Sept 2016 - Feb 2021

  - Member of the Adversarial Machine Learning lab. - Designed, implemented and published a signal processing-based two-level adversarial defense that detects the adversarial images produced to deceive deep learning-based image classification systems and transforms them back into the corresponding standard image samples. During my undergraduate studies in Computer Engineering, I had the chance to be the teaching assistant for the following courses:- Image Processing and its Applications- Introduction to Programming (C Language)- Linear Algebra and its Applications- Internet of Things- Computer Architecture- Research and Technical Presentation

  • Undergraduate Research Assistant

    Jan 2020 - Feb 2021

  • Undergraduate Teaching Assistant

    Sept 2016 - Sept 2020
• 

  BC Cancer

  Sept 2021 - now

  As a member of the High-Performance Computing team within the Bioinformatics Technology lab (Birol Lab) at Canada's Michael Smith Genome Sciences Centre, I actively participated in collaborative research over a span of two years.Underlying these efforts was the implementation and proposal of Stash, a novel hash-based data structure designed for storing large sequencing read datasets. Utilizing a two-dimensional structure based on hash values generated from spaced seed patterns, Stash compresses data efficiently. The representation of k-mers and sequence ID hashes enables comparisons to determine if queried k-mers are covered by the same set of sequences, extending the utility of Stash across bioinformatics domains. The novelty of Stash lies in its ability to detect and correct genome misassemblies based on signal processing techniques. In our extensive experiments, Stash reduces misassemblies while preserving assembly contiguity. This innovation holds promise for refining genomics research accuracy and reliability.Through this project, I had the opportunity to present my findings at prominent international conferences RECOMB2023 and ISMB2022, showcasing the significance of our work on a global stage. The project involved extensive signal processing, algorithm design, and code optimization. Show less

  • Bioinformatics Scientist

    Dec 2023 - now

  • Graduate Research Assistant

    Sept 2021 - Nov 2023
• 

  BC Children's Hospital Research Institute

  May 2022 - Jul 2022

  Bioinformatics Research Intern

  The projects I worked on during this internship are as follows:- Developing a Python software for generating allele frequency files using gnomAD allele frequency and family allele frequency. - Inferring the pedigree for a family enrolled in the Silent Genomes Project: the IBIS tool was used to call IBDs, and the 23andMe Bonsaitree software was utilized to infer the final pedigree. - Preparing long read analysis pipelines for PacBio HiFi and Oxford Nanopore sequencing reads. The pipelines include the following steps: Quality checking, Alignment, SNV/Indel calling, SV calling, Methylation calling, Short-tandem repeat genotyping Show less

• 

  The University of British Columbia

  Sept 2022 - Nov 2023

  Master of Data Science Teaching Assistant

  I have been the teaching assistant for the following courses:- Descriptive Statistics and Probability for Data Science,- Data Visualization I, - Data Visualization II, - Computing Platforms for Data Science- Communication and Argumentation in Data ScienceDuring the above courses, I delivered educational assistance to a large number of students through the facilitation of weekly lab sessions and consistent availability during office hours. I also evaluated and offered feedback on tutorial assignments, examinations, and project submissions, managing a workload encompassing as many as 50 students each week. Through personalized mentoring, I supported students in increasing their data science skills and enhancing their grasp of course contents. Show less

• 

  Vancouver General Hospital

  Mar 2023 - Nov 2023

  Machine Learning Engineer and Consultant

  As a Machine Learning Engineer at the Vancouver General Hospital Trauma Surgery Team, under the supervision of Dr. Morad Hameed, I have worked on data-driven analyses that impact patient care. My role has encompassed responsibilities such as the following:- Data Pre-processing: I have executed feature preprocessing techniques, such as scaling, imputation, one-hot encoding, and datetime feature analysis. I ensured that the raw data transforms into a structured dataset ready for predictive modeling.- Feature Selection: Employing methods like Recursive Feature Elimination, I have identified the most impactful features to empower our models. This approach helps the models to focus on the most relevant information.- Mitigating the Class Imbalance Problem: I have addressed class imbalance challenges by implementing undersampling and synthetic data generation techniques. These approaches ensure our models are not skewed towards the majority classes, enhancing their ability to generalize to real-world scenarios.- Outlier Detection: Leveraging techniques such as Local Outlier Factor and Isolation Forest, I have increased the robustness of our models against data anomalies and outliers. - Different Problems: I worked on both classification and regression models, such as K-Nearest Neighbors, ensemble models like XGBoost, LightGBM, CatBoost, and Adaboost, as well as techniques like Random Forest, SVM, SVR, and deep learning. This diversity ensures that the right tool is applied to the problem.- Optimization: Utilizing hyperparameter optimization and cross-validation methods, I have fine-tuned the models. This step also included assessing and adjusting model calibration to ensure that probabilistic predictions accurately reflect true likelihoods.- Progress Documentation: My dedication to documentation guarantees that our results are reproducible. - Collaboration: I have been an active participant in discussions and idea exchanges with the medical team. Show less