I’m Nima Sarang. I’m a Machine Learning Scientist at Expedia Group. I build and manage machine learning models for Search Engine Marketing (SEM), for Expedia Group’s global brands, like Expedia, Hotels.com and Vrbo.
I’m addicted to learning and sharing knowledge, and am enthusiastic about applying statistical and machine learning techniques to solve real-world problems.
About Me
Prior to working at Expedia Group, I was a Machine Learning Engineer at Divar, where I worked on pricing and computer vision models for the largest e-commerce platform in Iran. I had the joy and fortune of working with a lot of talented people, and learned how to build and deploy machine learning models at scale.
I completed my master’s degree in Computer Science at Concordia University, where I was a member of the Immersive & Creative Technologies Lab. I worked on applying deep reinforement learning to satellite imagery data for extracting road networks using controllable agents.
I’m a passionate about solving problems. In my free time, I often participate in Kaggle-like competitions to learn about applying ML techniques to different problems. I try to keep up with the latest research in the field, and am always looking for ways to improve my skills. One of the reason I started this website is to write about things I discover and share cool stuff.
Let’s do something awesome together!
If you’re interested in working with me, please drop me a line at [email protected]!
A little bit more about me…
Here is a timeline of my professional experience and the projects I’ve worked on:
-
2022 - Present
Expedia Group
Machine Learning Scientist
Capital Allocation
Capital optimization for search ads bidding on Expedia Group's global brands. My team manages a $100M+ annual budget.Sparsity-Aware ML Models
Novel ML, bayesian and tree-based methods for training on large-scale and highly sparse data.Real-time Controllers
Developed control-loop systems for real-time bidding on search engine marketing platforms. -
Sep 2021 - Mar 2022
Concordia University
Instructor
Ericsson ML/AI Upskill Training Program
I taught PyTorch and Computer Vision tutorials to Ericsson employees. I also mentored three teams throughout the program, guiding them through the implementation and debugging of their projects. -
2020 - 2022
Immersive & Creative Technologies Lab
ML Researcher
Tractable Large-scale Deep Reinforcement Learning
Leveraged deep reinforcement learning to solve massive-scale environments and developed an automatic extraction system for urban road networks from high-resolution aerial imagery, in collaboration with CAE. -
2021
Stock Trading Agent
I spent a good chunk of my free time developing a stock trading agent that uses reinforcement learning and forecasting models to make trading decisions. Even though it wasn't able to beat the B&H strategy, I learned a lot about trading, forecasting, scalability, and offline policy evaluation. -
2019 - 2020
Divar
Machine Learning Engineer
License Plate Detection and Anonymization
Developed a real-time pose estimation model for automatically hiding vehicle license plates in images, and published an educational technical blog on the implementation details. This is an example in action.Used Vehicle Price Estimation Model
Developed a used-car price valuation model that was deployed as a free SaaS to all users.Client-Side ML for Merchandise Valuation
Developed a client-side multi-task AI model for image classification and price estimation of merchandise and commodities in real-time. Deployed on Android using Java and TensorFlow Lite. Used fastText and TF-IDF to automatically tag unlabeled data. -
2019
Augmented Reality Soccer Using Deep Learning
As my Bachelor's thesis, I developed an two-player augmented reality soccer game played witha virtual ball and field. Built using Unity, an optimized semantic segmentation model, and an object tracking algorithm. -
2017 - 2018
Computational Biology Research Center
Research Assistant
Protein Design
Worked on designing protein sequences that can fold into a given tertiary structure using AI and evolutionary profiles. -
2016 - 2017
Amirkabir University
Competitive Programmer
ICPC - ACM International Collegiate Programming Contest
I was a member of the university's competitive programming team, and participated in the ACM ICPC regional contest. The algorithmic problems we solved varied from graph theory and dynamic programming, to computational geometry and greedy.