Ahsan Zaman, Developer in Los Angeles, CA, United States
Ahsan is currently unavailable

Ahsan Zaman

Software Developer

Los Angeles, CA, United States

Toptal member since May 10, 2022

Expertise
PythonAlgorithmsMachine LearningArtificial IntelligenceDeep LearningComputer VisionTensorFlowAgile DevelopmentPyTorchReact.jsJavaAWS
Bio

Ahsan is a software engineer with 5+ years building and scaling AI and cloud platforms. At Amazon, he architected production ML pipelines using SageMaker, EMR, and Step Functions to improve ad relevance across 100+ million daily transactions, driving $90+ million in annual revenue. At Northrop Grumman, he led computer vision and reinforcement learning systems for defense applications. Ahsan specializes in model training and orchestration, real-time applications, and A/B experimentation.

Portfolio

Amazon Advertising
Agile Software Development, Amazon Web Services (AWS), APIs, AWS Deployment
Northrop Grumman
Machine Learning, Computer Vision, Python, Image Recognition...

Experience

  • Algorithms - 5 years
  • Python 3 - 5 years
  • TensorFlow - 4 years
  • Machine Learning - 4 years
  • Agile Software Development - 3 years
  • React - 3 years
  • Computer Vision - 2 years
  • PyTorch - 2 years

Preferred Environment

Python 3, PyTorch, TensorFlow, MacOS, Linux, Python, Amazon Web Services (AWS)

The most amazing...

...system I've built is a production ML pipeline at Amazon improving ad relevance for 100+ million daily queries, driving $90+ million in annual revenue.

Work Experience

Software Development Engineer II

2022 - PRESENT
Amazon Advertising
  • Designed multi-stage AWS Step Functions workflows integrating SageMaker training jobs, EMR processing, and Redis cache publishing to deploy and validate ad relevance ML models in production.
  • Improved system reliability by introducing proactive alerting and on-call dashboards, achieving 99.75%+ uptime across high-traffic advertising infrastructure.
  • Led full-stack architecture of scalable AWS-based APIs and web pages for a high-traffic advertising platform handling 100+ million daily API calls, overseeing React front-end and Java/Python microservices.
  • Implemented AWS EMR big data pipelines for product ad recommendations integrated into real-time recommendation systems, increasing recommendation coverage by 8%.
  • Built end-to-end Advertiser Console pages using PySpark on EMR, APIs on AWS Fargate, and React/TypeScript on CloudFront, generating $90+ million in annual attributed revenue.
  • Designed an extensible cache automation system using AWS Step Functions and EventBridge to refresh auction signals for Sponsored Products ads, eliminating days of manual engineering effort.
  • Architected event-driven, serverless data pipelines processing 100+ million daily transactions using AWS Lambda, Step Functions, and DynamoDB at production scale.
Technologies: Agile Software Development, Amazon Web Services (AWS), APIs, AWS Deployment

Software Engineer, Artificial Intelligence

2020 - 2022
Northrop Grumman
  • Led technical efforts for the computer vision sub-team. Researched the latest advancements in computer vision as relevant to my team's statement of work, proposed new projects, and guided team members and interns.
  • Developed computer vision classifiers for synthetic image data customized to my team's specific needs. Achieved over 98% accuracy in object detection tasks.
  • Applied deep-Q learning and soft actor-critic algorithms to create reinforcement learning controllers for simulation platforms requiring both discrete and continuous control.
  • Ensured smooth deployment of projects via Docker containerization.
Technologies: Machine Learning, Computer Vision, Python, Image Recognition, Reinforcement Learning, Artificial Intelligence (AI), Deep Learning, Keras

Experience

End-to-end Computer Vision Pipeline

Developed a pipeline for generating synthetic images, training object recognition classifiers on those images, generating metrics, and exporting chosen classifiers to deployment environments using Docker.

Reinforcement Learning Controller for Simulated Platforms

Developed continuous and discrete controllers for simulation platforms, achieving the optimal simulations result in over 98% of test scenarios. Algorithms used include variants of deep-Q networks and soft actor-critic reinforcement learning algorithms.

Education

2019 - 2020

Master's Degree in Computer Science

University of Southern California - Los Angeles, California, United States

2015 - 2019

Bachelor's Degree in Computer Engineering and Computer Science

University of Southern California - Los Angeles, California, United States

Skills

Libraries/APIs

PyTorch, TensorFlow, Keras, React

Tools

You Only Look Once (YOLO), AWS Deployment

Languages

Python 3, Python, Java

Paradigms

Agile Software Development

Platforms

Amazon Web Services (AWS)

Other

Machine Learning, Computer Vision, Algorithms, Deep Reinforcement Learning, Artificial Intelligence (AI), Image Recognition, Reinforcement Learning, Deep Learning, Physics, APIs

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring