Timur Usmanov

Experience

• 

  Periplus AG

  Jun 2019 - Aug 2019

  Machine Learning Intern

  Participated in the internship program under the supervision of experienced mentors. Learned the basics of Artificial Intelligence, and Machine Learning specifically in Python (using Tensorflow/Keras, Pandas, Numpy, scikit-learn).

• 

  LDKR Group, Inha University

  Feb 2020 - Feb 2021

  Researhcer

  Participated in a team to create an analogue of Princeton WordNet for further use by applications in Uzbek language:- Analyzed the structure of PWN and the source files- Performed data filtering and formatting of database files using Python- Used a lexical resource (dictionary) for automatic translation of PWN concepts- Optimized the time to query the database for faster structure traversal

• 

  "Single integrator - UZINFOCOM"​ LLC

  Sept 2020 - Mar 2022

  Python Developer

  𝗠𝘆𝗜𝗗URL: https://myid.uz/en/Contribution:- Played a key role by developing an AI solution for biometric authentication to extend remote services portfolio of leading banks and marketplaces of Uzbekistan.- Researched the available liveness detection and face recognition AI models, built and trained (using Tensorflow) a liveness detection model that achieved a FPR of less than 2%.- To serve the models, created a service with FastAPI and Docker, and integrated the service with the rest of the idetification system. This allowed for automation of user verification allowing for significant cost reduction via employing less staff for manual application processing.- Developed new features for and maintained the main identification service in Django, improving user experience.𝗣𝗶𝗶𝗺𝗮𝗜𝗗Contribution:- Created an authentication service with Flask and Docker to enable face recognition in authentication flow for Presidential Schools, which allowed to engage 2 times less school staff during the entrance exam check-in process.- Sped up registration of school staff and applicants by creating a NoSQL (Apache Solr) database for vectors for fast face searching and flexible profile creation and modification.- Reduced the time to process creation of new profiles and check-ins by 2-3 times via utilizing capabilities of Tensorflow for AI model inference on multiple GPUs. Show less

• 

  EPAM Systems

  Apr 2022 - Oct 2024

  Software Engineer

  𝗗𝗮𝘁𝗮 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗼𝗳 𝗮 𝗯𝗶𝗴 𝗠𝘂𝗹𝘁𝗶𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗖𝗼𝗻𝗴𝗹𝗼𝗺𝗲𝗿𝗮𝘁𝗲Contribution:- Led the implementation of new features for part of the platform - broke down the requirements and collaborated with colleagues-experts in Snowflake, coordinated delivery of features, discussed potential design and integration with the frontend team.- Managed deployments, enforced good practices: code coverage of 90%, SOLID.- Created new serverless API using combination of API Gateway and Lambda functions, used Terraform configuration for IaaC management. The API allowed to automate management of isolated user spaces in Snowflake, eliminating the burden of processing user requests from experienced data engineers on the platform.- Improved user experience using optimizations for AWS Lambda functions. Maintained existing APIs in AWS environment (API Gateway, Lambda, RDS).𝗦𝗲𝗻𝘀𝗶𝘁𝗶𝘃𝗲 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗟𝗲𝗮𝗸 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗦𝘆𝘀𝘁𝗲𝗺 (𝗘𝗣𝗔𝗠 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝗹 𝗣𝗿𝗼𝗷𝗲𝗰𝘁)Contribution:- Developed a service for leak detection automation (PII, access keys) in git repositories using FastAPI and Docker in AWS environment. Enhanced flexibility and scalability by refactoring the service into 3 separate ones, used SQS for communication.- Subscribed the service to Kinesis topic that was pushing commit events. Integrated the service with Github API for fetching commit information and code diffs, and employed an S3 bucket to store commit information in order to reduce calls to Github API for the repositories recurring in Kinesis messages.- Integrated the service with AI pipelines (SageMaker) to verify the potential leaks, which has lowered the number of false positive findings by more than 60%, which allowed for lesser utilization of humans for verification of the potential leaks. Show less

• 

  Yandex

  Oct 2024 - now

  Software Engineer