2023 ? 2024: counter-drone hardware (detection and jamming)

Customer: Miltech startup

Tasks:

Skills:

GNU Radio, SDR, NumPy, SciPy, Embedded Linux, Orange Pi, UART

2020 - 2023: Zoral

Role: Python Backend Engineer

Customer: fintech

Tasks:

• Developed scalable backend services focused on PDF processing, AI integration, and AWS automation.
• Built and optimized processing pipelines for large batches of financial documents - multi-fold performance improvements, thousands of pages in 5 minutes.
• Mixed quality: digital text, raw scans, broken fonts, broken PDF structure. If extract is garbage, rasterize + OCR and add a text layer.
• OCR of scanned pages, page cropping and filtering, compression and optimization of embedded images, overall file size reduction without loss of quality.
• Cut handling from 40 minutes plus manual checks to 5 minutes.
• Designed REST APIs and asynchronous workflows using Celery and RabbitMQ.

Skills:

Python, Flask, Celery, RabbitMQ, Ghostscript, PyMuPDF, ImageMagick, AWS (Lambda, S3, Textract)

2015 ? 2019: Machine Learning

Role: Python Engineer

Customer: Lohika

Tasks:

• Bots / NLP:
  • customer-specific NLP bots; train and deploy ML apps to cloud (Flask, Celery, Airflow, Kubernetes, Docker).
• Distributed training:
  • web APIs for third-party ML models; distributed training system; deployment and performance work (Flask, Kubernetes, Docker).
• Media assets:
  • face detection/recognition, image tagging, NSFW/inappropriate-content detection (OpenCV, Keras, TensorFlow, dlib, Docker).
• ITraffic:
  • empty parking-lot detection from webcam ? datasets, models, API, deployment (OpenCV, Caffe, Docker, RabbitMQ, AKAZE/ORB/BRISK, homography).
• Autonomous robocar:
  • control, navigation, mapping for a small 4W robot; SLAM experiments, hardware bring-up (OpenCV, ROS, LIDAR, stereo camera).
• Gene variations:
  • classify pathogenic gene variants (pandas, scikit-learn).

Skills:

Flask, Celery, Airflow, Kubernetes, Docker, OpenCV, Keras, TensorFlow, dlib, RabbitMQ, AKAZE/ORB/BRISK, homography, ROS, LIDAR, stereo camera

2007 ? 2015: various Projects

Role: Java / Python engineer, team lead

Tasks:

• Crawling / scraping - large-scale web crawling and data extraction.
• Android / Java - client apps and team lead.
• Java backends - product and outsourcing work.

Skills:

Python, RabbitMQ, MySQL, PhantomJS, Redshift, S3, EC2, Elasticsearch, NLTK, Java, Android 2.x, Spring, Hibernate, Maven, Gradle, Scala

2018 ? 2019: Image search engine

Tasks:

• AI-powered image search engine with facial recognition, object detection, image classification, and metadata extraction.
• Search by plain text descriptions; filter by tags, metadata, faces, composition.
• Nomad:
  • Distributed crawler (TitanDB, OrientDB, Gremlin).
• Firefly:
  • Web manager for scientific papers (GridFS, TitanDB, Elasticsearch).
• Denigma:
  • open-source participation
    • Ageing-research tooling: backend, knowledge extraction, some ops (CloudStack, Scala, NLTK, OpenNLP).
• BigPlanet
  • Offline maps with multiple geo data sources (Android).

Stack:

Python, Flask, RabbitMQ, MongoDB, TensorFlow/Keras, Docker, Kubernetes, React, TitanDB, OrientDB, Gremlin, GridFS, TitanDB, Elasticsearch, CloudStack, Scala, NLTK, OpenNLP, Android

various Projects

Customer: on request

Tasks:

• Anki-style flashcards for learning words: from a list, a marked photo, or a topic -> German lemmas, English glosses, examples.
• OCR + LLM pipeline for vocabulary extraction and generation.

Skills:

Django, Docker, OpenAI