Akash Venkateshwaran Portfolio

I am Akash Venkateshwaran

Name: Akash Venkateshwaran

Email: akash.venkateshwaran@gmail.com
akashv22@student.ubc.ca

About me

I am a passionate research engineer dedicated to building data-driven solutions that solve complex engineering and business problems. My journey began in computational physics, where I developed a deep foundation in numerical analysis and mathematical modeling to understand and solve fundamental problems in physics. This experience ignited my passion for creating simulation worlds and predictive models, particularly for industries where precision matters most—marine, environmental, and nuclear sectors.

My expertise has evolved to focus on scientific machine learning (SciML), where I develop predictive models that seamlessly combine domain knowledge with state-of-the-art ML techniques. I have extensive experience building models from both real-world datasets and numerical simulations—ranging from quantitative analyses and time-series forecasting to regression approaches. My research also encompasses operations research, where I’ve designed decision-support systems for maritime operations. Applying advanced optimization methods, I’ve enhanced decision-making processes across a range of complex engineering scenarios.

I am continually driven to explore new methodologies and expand my skill set at the intersection of data science, machine learning, and operations research. If you’re interested in collaborative research to address industry-scale challenges, I would welcome the opportunity to connect and develop impactful solutions together.

Education

University of British Columbia

2022 – 2025

Vancouver, Canada

M.A.Sc. in Mechanical Engineering, Grades: 90.2%

Thesis:A Decision Support System for Minimizing Underwater Radiated Noise from Ships
Supervisor: Prof. Rajeev Jaiman
Coursework: Computational Optimization, Machine Learning, Industrial Robotics and Deep Learning with Graph
Honors and Awards: Mitacs Globalink Graduate Fellowship, International Tuition Award, Academic Achievements Award

Vellore Institute of Technology

2018 – 2022

Chennai, India

B.Tech in Mechanical Engineering, Grades:96.2%

Capstone Project: Numerical study of heat transfer characteristics in an internally heated meltpool due to forced convection
Supervisor: Prof. Shyam Kumar M. B
Honors and Awards: Achieved 5th rank in the mechanical engineering department

Experience

Research Engineer

September 2022 – April 2025

Computational Multiphysics Laboratory (CML) - University of British Columbia, Vancouver, Canada

Collaborated with Clear Seas and marine consultation company JASCO to develop a Decision Support System (DSS) aimed at reducing ship noise signature footprints in the Salish Sea. Developed a pioneering predictive model achieving high SSIM accuracy for far-field acoustic modeling through a GPU-enabled Conditional Convolutional Neural Network (RC-CAN) trained on UBC ARC Sockeye HPC clusters. Enhanced model generalizability using replay-based training strategies and performed large-scale numerical simulations of acoustic propagation, generating a comprehensive relational database of acoustics data. Engineered robust data ingestion and preprocessing pipelines for time series meteorological data from NOAA stations and developed automated data pipelines to streamline simulation processes, enabling parallel execution to enhance computational efficiency. Built an end-to-end DeepAR forecasting model for systematic model versioning, experiment tracking, and production deployment workflows, implementing rigorous backtesting frameworks and systematic hyperparameter optimization using Optuna to ensure model reliability and performance. Modeled route and speed optimization problems focused on reducing ship noise signatures, investigating meta-heuristic, graph-based, and sample-based search algorithms with Pareto pruning and constraint-handling techniques. Developed an interactive simulation environment featuring dynamic marine mammals and AIS-based ship voyages to analyze underwater noise footprints, successfully demonstrating significant reduction in noise exposure. Integrated the RC-CAN model into the DSS for real-time 3D noise signature mapping and ship route optimization, supporting diverse scenario testing and operational insights.

Data Science Intern

September 2024 – December 2024

IPEX Technologies Inc. - Mississauga, Canada

Implemented multiple Gaussian Process Regression (GPR) models within Hex's workspace integrated with Snowflake to predict polymer material properties based on recipe ingredients and processing conditions. Developed robust data pipelines for exploratory data analysis and feature engineering within Hex, creating interactive display cells to optimize the representation of key variables. Achieved significant improvement in predictive accuracy across various material properties, surpassing baseline models offered by Uncountable and demonstrating enhanced performance over naive modeling approaches.

Software Engineer

September 2023 – September 2024

UBC Sailbot - Vancouver, Canada

Led the optimization team, formulating the optimization problem, identifying the design variables and constraints, and implementing software for an autonomous sailboat. Developed a decision-making algorithm leveraging sensor data fusion for autonomous navigation of sailbots. Created various test cases, performance benchmarks, and convergence studies to assess different optimization algorithms.

Mitacs Globalink Research Intern

June 2021 – August 2022

Environmental Hydro-Dynamics (EHD) Laboratory - York University, Toronto, Canada

Simulated surge waves in multiphase environments using OpenFOAM, focusing on turbulent structures and vortex dynamics. Developed statistical analysis algorithms for Reynolds stress tensor and conducted comparative simulations. Documented and published findings in journals and conferences.

Research Intern

September 2020 – March 2021

Integrated MechanoBioSystems Lab - National Cheng Kung University, Tainan, Taiwan

Developed a cervical cell segmentation model using Mask R-CNN, achieving high classification accuracy. Utilized deep learning for segmentation of cell regions, achieving over 0.95 accuracy with low variance.