Daniyar Bakir Email & Phone Number

- Replaced and structurized old system of .docx and .xlsx files by designing and building DB, front- and back-end systems of the WebApp- Improved CICD completion time by 2 times and reduced image size by 4 times via optimizing Dockerfile, CICD pipeline and introducing more sophisticated package system- Designed and developed the web scraper and report generation system that replaced the previous manual system of data search and report preparation

Nur-Sultan, Kazakhstan

Working on Speech-to-Text Technology (STT)Worked on:- Audio Data Annotation Pipeline- STT Model Training- STT Deployment- Audio, Text and Image Data CollectionAchieved:- Fully functional STT service- Iterational improvements of STT model- Automation of Data Preparation Pipeline via Airflow- Implementation of some steps of Audio Annotation Pipeline- Audio Annotation Web Tool- Social-Networks crawling system based on message-queue architectureResponsibilities:- Research and development of STT models- Development of Audio Annotation Pipeline- Configuration and optimization of training recipes- Report of technical and business metrics- Choose STT model for different tasks methodologicallyResults:- Training scripts for Kazakh-Russian languages environment- Easy to move and setup training receipts with different STT architectures- 30% reduction in time consumption of Spoken Language Identification Step in Audio Annotation Pipeline- 6 hours reduction of the STT model training time by optimizing feature preparation stages- Freeing up GPU server occupation by 1 day via separating feature preparation/CPU computation stages from GPU training steps- Optimized time consumption of data preparation pipeline by 60% by restructuring workers into a distributed system

Astana, Kazakhstan

Worked On:- Assistant to ProfessorsResponsibilities:- Prepare lecture and laboratory rooms and required equipment for academic process- Conduct Labs and Tutorials- Assess and supervise students' performance

Astana, Kazakhstan

Worked on:- High-dimensional classifiers- New Ranking System of UniversitiesAchieved:- Efficient estimation method of optimal regularization parameter of Regularized Linear Discriminant Analysis (RLDA) and publication of that method (https://pubmed.ncbi.nlm.nih.gov/27485443/)- Extensive comparison of high-dimensional classifiers under different configurations of sample and feature size ratios- Data crawling system of academic data based on self-constructed message-queue architecture- Full development and deployment of web-site to visualize the results of Universities Ranking SystemResponsibilities:- Development, coding and extensive comparison tests for efficient estimation method of regularization parameter in RLDA and other high-dimensional classifiers- Development of the crawling system of academic data- Back- and front-end development of the web-site on the Universities Ranking SystemResults:- 10 fold time increase in estimating the optimal regularization parameter of RLDA with comparable classification metrics to conventional cross-validation estimation techniques- Crawled academic data for 5000+ universities from 4 sources and further matching of data across different data sources- Fully functional web-site of academic ranking system (https://a3ranking.com)

Astana, Kazakhstan

As an assistant of Chief Electrical Engineering in a Construction company my duties and responsibilities included:- Designing electrical system within and outside of the building (power estimation, AutoCAD drawing, equipment selection)- Cost and installment workload estimation of the designed electrical system- Verification and validation of previously the designed electrical systems

Master’S Degree, Electrical And Electronics Engineering

Bachelor’S Degree, Electrical And Electronics Engineering

Paper (Capstone Project) titled "Activity Detection with Dendrite Threshold Model", International Conference on Advances in Computing.