Email: mark.vinogradov@phystech.edu
I am an experienced software engineer with broad ML background. I’ve designed and implemented several scalable SaaS products. My preferred language is Python, although i am familiar with Kotlin, Java and Golang.
My strongest quality as an engineer is that I can execute software projects end-to-end, including everything what is expected from industrial-quality product: architecture designs, neat and readable code, extensive multi-level testing, documentation, deployment and monitoring. As a bonus, my machine learning background helps tackle common challenges like train set annotation, model versioning, bias mitigation, scaling and performance monitoring.
Outside of work, i dedicate most of my time to my family, but I also enjoy designing new quirky tabletop games.
Contributed to both backend (web app, async CPU-heavy task
scheduler, other microservices) and ML submodules (yield
productivity maps, yield data analysis, soil sampling).
Designed and implemented various other features (billing &
monetization, monitoring, data anomaly detection etc.).
Developed Python & JVM-powered backend for SaaS cloud platform (GCP-hosted) providing ML model inference through a public API.
Achievements:
Developed web app service with on-premise deployed data
collection agent.
Led a team of 3 junior-level engineers to build a backend
for medical data collection & processing.
Worked on PoCs for improving object detection &
classification on the retail store photos.
Integrated the improvements as features for backend service
and detection & classification service.
Led the project on automatic shelf monitoring with
X5.ru.
EDH Pairings: A tournament management system for the competitive Magic: the Gathering EDH format in a four-man groups of players. Designed as a Django service with password-less authorization, server-side page rendering with REST API and simple mobile-friendly design. GitHub
Shooter: full-page site screenshot service: REST API and CLI-enabled tool that captures full-page or partial screenshots from URLs, detects and labels HTML elements, and performs scripted user actions during the process. Designed a scalable architecture with task brokering, enabling concurrent execution by multiple workers and delivering comprehensive outputs, including labeled images and JSON metadata. GitHub
Master in Computer Science, Skoltech Institute of Science and Technology (2017 - 2019)
Thesis: Comparing the cost efficiency of image annotation task sets. [pdf]
Bachelor in Computer Science, Moscow Institute of Physics and Technology (2012 - 2016)
Thesis: Comparison of trend extraction methods for anomaly detection. [pdf (ru)]