cv

Mid-Level Data Scientist

Tapsi

As a member of the Data Science, Maps team, I contributed to an in-house ETA prediction system. This involved processing GPS data from drivers, forecasting traffic speeds on road segments, determining optimal routes, and accurately predicting trip ETAs.

• Unified backend and data science ETA modules into a centralized repository.
• Architected a graph-based ETA prediction system with Graph Neural Networks.
• Enhanced preprocessing pipelines with seasonality-aware adjustments.
• Optimized location matching with improved OSRM routing engine algorithms.
• Automated model retraining and deployment for the ETA framework.

Mid-Level Software Engineer

Tapsi

As a member of Backend, Rides team, I worked on the core ride-hailing platform, focusing on driver-passenger matching, ride dispatching, and driver-location tracking.

• Developed a flow-optimized driver-passenger matching algorithm in collaboration with the data science team.
• Refactored microservices by optimizing database queries, indexing, and caching strategies, and migrated inter-service communication to gRPC.
• Created monitoring dashboards for microservices using Grafana and Metabase.

Scientific Committee Member

Young Scholars Club (YSC)

Lecturer and Examiner for IOAA & INOAA Team Selection.

• Lecturer and Examiner
• Team Selection for IOAA & INOAA

B.Sc

Sharif University of Technology

Computer Engineering

• Deep Learning (20/20)
• Machine Learning (20/20)
• Artificial Intelligence (20/20)
• Advanced Information Retrieval (20/20)
• Linear Algebra (19.4/20)
• Probability and Statistics
• Computer Simulation
• Game Theory

Silver Medal

International Olympiad on Astronomy and Astrophysics

Gold Medal (1st Place)

Ministry of Education, Iran

Best Astronomical Observer

Astronimical Society of Iran

1st Place in the Messier Marathon Competition.

Best Astronomical Observer

Ava Star Astronomy Center

1st Place in the StarCup Competition.

Unlabeled Out-of-Domain Data Improves Generalization

International Conference on Learning Representations (ICLR), 2024

Vision-Language Models Toolbox

A Python library for fine-tuning and evaluation of vision-language models such as CLIP & BLIP with PyTorch, supporting various datasets and tasks like image classification.

MedSegDiff: Medical Image Segmentation with Diffusion Model

Implements the MedSegDiff: Medical Image Segmentation with Diffusion Probabilistic Model paper for medical image segmentation, using the LGG Segmentation Dataset to detect tumor and cancer anomalies.

Bootstrap Your Own Latent (BYOL)

A Pytorch implementation of BYOL using a pre-trained ResNet backbone on the STL10 dataset.

Generative Models

Experiments with VAEs, GANs, and DDPMs, on the Fashion MNIST dataset.

Zero-shot Object Detection

Zero-shot object detection with CLIP, utilizing Faster R-CNN for region proposals.

Adversarial Robustness

Evaluating robustness against adversarial attacks such as FGSM and PGD.

Full-stack Food Delivery Platform

A food delivery application built with a Python backend (FastAPI & Docker) and a Vue.js frontend. It presents an event-driven microservices architecture with many modern features!

FastAPI JWT Authentication

A scalable authentication middleware for FastAPI, employing Redis to provide efficient JWT-based authentication and seamless integration with distributed systems.

Messaging & DBMS Libraries

Python libraries for distributed event-driven architectures (rabbitmq-rpc) and streamlined database interactions and connectivity (mongo-motors, asyncpg-client, aredis-client).