Skip to content
MZ

Margulan Zholaman

@Margulan

Software Engineer, MSc in Computer Science | Python, Django | Distributed Systems | AWS

Software Engineer, MSc in Computer Science | Python, Django, FastAPI | Distributed Systems | AWS | Go | Docker |

Career

Experience: 2 years

Experience

Backend Developer

State Technical Service

Sep 2024 — Present

Built a high-load real-time chat and improved deployment efficiency and API response time.

• Developed real-time chat with 1000+ concurrent users by using Django Channels with Redis for tracking online presence, maintaining maximum 50ms latency and stable message delivery during high-load simulations • Reduced Docker image size by 40% by optimizing Dockerfiles by multi-stage builds, .dockerignore file, and slim base images - resulting in improving deployment speed and CI pipeline performance • Decreased average API response time by 30% by configuring Nginx as a reverse proxy with gzip compression and implementing Redis caching for frequently requested endpoints

Backend Developer

Sparrow

Sep 2023 — Sep 2024

Built financial-market backend services processing over 7,000 tickers daily and improved request performance.

• Developed a backend service that processes over 7,000 tickers daily, optimizing real-time data handling for global financial markets • Improved system performance by reducing average response time for high-load requests by 13% through database query optimization • Improved application responsiveness by integrating Celery, providing background task processing for SMS notifications, stock data collection, and trader statistics calculations

Research Assistant

Nazarbayev University

Oct 2023 — Dec 2023

Implemented energy-saving CNN training techniques, reducing training energy consumption by 83% and processing time by 30%.

• Implemented techniques from the research paper titled ”E2-Train: Training State-of-the-art CNNs with Over 80% Less Energy.” • Implemented Stochastic mini-batch dropping, Input-dependent selective layer update, and the Predictive sign gradient descent approaches, to reduce energy consumption of model training by 83% • Utilized CUDA and PyTorch for computationally intensive tasks, achieving a significant 30% reduction in processing time compared to conventional methods

Education

M.S., Computer Science

Nazarbayev University

2024 — 2026

B.S., Computer Science

Nazarbayev University

2020 — 2024

Selected projects

Thesis RAG Analyzer

Designed and deployed an event-driven RAG pipeline on AWS, using S3 event notifications and SQS to decouple document ingestion from a Go-based query API running on ECS, enabling independent scaling of ingestion and query workloads. Built an embedding worker on EC2 that processes uploaded thesis PDFs and indexes them into a vector database, with idempotent writes to prevent duplication from SQS’...

Skills

PythonGoSQLDjangoDockerKubernetesFastAPILinuxGitKafkaRabbitMQNginxPostgreSQLRedisMongoDBAWS (ECS, EC2, S3, SQS, IAM)Terraform