Pedram Agand

Experience

• 

  ASR KISH Drilling company

  Jun 2013 - Aug 2013

  Electrical Engineer

  supervisor in SCR room to evaluate the required DC Voltage that was Supplied by Diesel Generator KATERPILLAR

• 

  Respina Networks & Beyond

  Aug 2016 - Dec 2016

  Research And Development Specialist

  Develop Billing system for VoIP with Python

• 

  Fara Herfe Pendar

  Aug 2017 - Aug 2019

  Data Scientist

  Data scientist in smart home design.Provides smart home solutions for home and offices. The two key points are the outstanding solutions with the lowest cost for the customers. Smart home packages include lightening, curtain, security, air conditioning, irrigation, multi-media and other on-demand facilities namely remote access, scheduling, integrating and comfortably.Utilize data from users interactions to find the most optimum solutions to new costumers.

• 

  Simon Fraser University

  Sept 2019 - Sept 2021

  Research Assistant

  - Human navigational intent prediction using probabilistic approaches- Advance recursive MCMC approach for soft contact environment prediction - DMODE: Differential Monocular Object Distance Estimation Estimating object’s distance by alternation in object’s size over time via a single camera and IMU.- Distilled Multi-Task Learning for Transformer-Based Sensor FusionDMFuser uses multiple RGB-D for semantic segmentation and depth perception to generate end-to-end navigational commands with an attention-CNN and hybrid control in Carla Show less

• 

  Huawei Technologies Canada Co., Ltd.

  Jan 2020 - Dec 2020

  Research assistance (Joint project with SFU)

  Human states (position and heading) prediction using bayesian and optimal approaches.

• 

  Breeze Traffic

  May 2021 - Aug 2021

  Applied Scientist

  Propose EcoLight, as an efficient reward shaping for traffic signal control with minimizing co2 emission as the primary goal.

• 

  National Research Council Canada / Conseil national de recherches Canada

  Sept 2021 - Aug 2024

  Research Assistant (Joint project with SFU)

  Utilizing machine learning algorithm to optimize the fuel efficiency of BC ferry- Preprocessing data, and data wrangling (handling corrupted/missing entries, outlier detection, data normalization, clustering for homogeneous data, )- Employing domain knowledge in the design process of gray-box models and feature selection- Enabling statistical study to assess the data before applying prediction- Developing parametric/non-parametric models for fuel efficiency- Utilizing Ensemble techniques (bagging/boosting) e.g. XG-boost, Adam-boost, gradient-boosting- Employing deep networks to encapsulate mutual and time dependency- Transformer-based forecasting model with Gym simulator and D4RL-compatible dataset - True offline RL with ReBRAC optimizing navigational practice scheduling and fuel consumption Show less

• 

  Da-Tu

  May 2022 - now

  ML Applied Scientist

  - Explanatory data analysis with Streamlit (Dev. to Dep.), recommendation system in AI-tutor - Sequential models (DistilBert) for score prediction, enhancing decision-making accuracy by 25%. - AI-assisted recommendations that instruct users according to their interaction logs with Local LLM (Llama 3), a vector DB of rules, and LangGraph. The deployed microservice API improves users' scores by 31%.

• 

  Borealis AI

  Sept 2022 - Dec 2022

  Machine Learning Researcher

  TransDR: transformer based Bi-directional density ratio estimationIntroducing modified KL divergence loss by adding square loss and bi-direction computation to enhance training.

• 

  Vancouver Coastal Health

  Apr 2024 - Jul 2024

  Machine Learning Engineer

  Fine-tuned self-corrective RAG agent using Phi3, QLoRA, and Langgraph to extract information from meeting notes/transcripts, reducing staff data collection time by an average of 6 hours weekly.