Samen

Semen Bondarenko
Python Backend Engineer | AI/ML Engineer
[відкрити контакти](див. вище в блоці «контактна інформація») | [відкрити контакти](див. вище в блоці «контактна інформація») | Ukraine
LIVE PORTFOLIO: samen-bondarenko.com | PUBLIC PROJECT GITHUB: github.com/SamenB/SBGallery

PROFESSIONAL SUMMARY
Python Backend & AI Engineer with 3+ years of commercial experience building production FastAPI backends, data-driven
services, and applied AI/ML systems. Experienced with PostgreSQL, Redis, Celery, Docker, CI/CD, secure auth, payments,
webhooks, and backend architecture. Built and launched SBGallery as a live commercial platform, and contributed to ML/AI
projects across Computer Vision, signal analytics, medical imaging, LLM/RAG pipelines, vector search, and AI-agent workflows.

TECHNICAL SKILLS
Backend: Python, FastAPI, Pydantic v2, SQLAlchemy 2, AI/ML & LLM Systems: PyTorch, TensorFlow/Keras,
Alembic, PostgreSQL, asyncpg, Redis, Celery, REST APIs, Computer Vision, NLP, Transformers, transfer learning, fine-
JWT, OAuth2, webhooks, async/await tuning, RAG, vector search, ChromaDB, pgvector,
LangChain, LangGraph, PydanticAI, agent workflows

Architecture & Quality: Service Layer, Repository Pattern, Infrastructure: Docker, Docker Compose, Nginx, GitHub
Unit of Work, Data Mapper, provider adapters, domain Actions, CI/CD, Linux, TLS, Prometheus, Grafana, structured
exceptions, transaction management, caching, rate limiting, logging, automated backups
pytest, Ruff

EXPERIENCE

SBGallery - Founder & Solo Backend/Full-Stack Engineer 09/2025 - Present
Live commercial art-commerce platform for original artworks and print-on-demand products, with real payments, order
processing, admin tooling, and worldwide fulfillment. Live: samen-bondarenko.com | GitHub: github.com/SamenB/SBGallery

• Architected and launched a production FastAPI commerce backend with Python 3.12, async SQLAlchemy, PostgreSQL,
Redis, Celery, Alembic, service/repository layers, Unit of Work transactions, domain exceptions, and background jobs.
• Built secure authentication, authorization, and checkout flows: JWT access/refresh tokens in HTTP-only cookies, Google
OAuth, Argon2 password hashing, Redis token whitelist/blacklist, rate limiting, server-owned pricing, cart/order lifecycle, and
abandoned-order cleanup.
• Integrated real payment and fulfillment infrastructure: Monobank acquiring, ECDSA webhook verification, Prodigi Print API,
provider-adapter boundary, fulfillment preflight checks, idempotency keys, retries, callbacks, and order/status
synchronization.
• Designed a print-on-demand catalog pipeline: raw CSV ingestion, product curation/parsing, country-aware pricing and
routing, storefront payload materialization, print asset generation, S3 upload/verification, and admin tools for product
readiness.
• Delivered production operations and quality: 12-service Docker Compose stack, Nginx with TLS/HTTP2/security headers,
GitHub Actions CI/CD, Alembic migrations, Prometheus/Grafana monitoring, structured logs, automated PostgreSQL/media
backups, and 260+ backend tests.

Freelance / Contract - Python Backend & AI Engineer 2025 - Present
Practical LLM/RAG and AI-agent features for recruitment and CMS/e-commerce workflows, focused on backend APIs,
embeddings pipelines, vector search, prompt logic, and local evaluation/testing.
• CV-to-vacancy analysis service: keyword extraction for search, vacancy parsing/ranking, relevance scoring against a resume, and short
AI-generated comments for decision support.
• RAG backend for product knowledge: large CSV ingestion, chunking, embeddings generation, ChromaDB persistence, retrieval logic,
and AI-widget endpoints for product Q&A.
• Backend integration: embedding refresh endpoints, data validation, prompt templates, structured responses, FastAPI APIs, OpenAI-style
LLM APIs, and retrieval-quality testing.

Absolutist - ML Engineer & Backend Engineer 07/2023 - 09/2025
Worked in a ML outsourcing/R&D team and backend product team, delivering commercial PoC/MVP-stage machine learning
modules and web systems.
Selected commercial ML / AI projects:
Medical Imaging AI / Hand & Wrist Fracture Detection
Commercial medical CV project for assisted fracture screening in hand/wrist X-ray studies, focused on phalanges, metacarpals,
wrist zones, distal radius, and ulna regions.
• Built an end-to-end PyTorch pipeline for X-ray fracture detection: DICOM/image loading, OpenCV/NumPy preprocessing,
intensity normalization/windowing, anatomical ROI crops, Albumentations augmentation, fold validation, checkpointing,
metric tracking, and threshold tuning.
• Implemented segmentation-assisted region extraction using UNet-style models with timm backbones, mask/heatmap-guided
crops, consistent train/inference transforms, fallback crop handling, and region-level datasets for fracture classification.
Page 1
Semen Bondarenko

EXPERIENCE - CONTINUED

• Developed 2.5D-style classification experiments with EfficientNetV2/ConvNeXt backbones, multi-channel ROI inputs, CNN
feature extraction, LSTM / Transformer-attention aggregation, mixed-precision training, weighted BCE/focal-style losses,
mixup, and class-imbalance handling.
• Evaluated models with KFold/StratifiedKFold validation, augmentation/loss/backbone comparisons, precision-recall
threshold calibration, and inference outputs including fracture probability, suspicious region metadata, confidence score, and
JSON-ready API payloads.

Industrial Sensor Analytics / Predictive Maintenance Platform
Commercial ML module for an industrial IoT condition-monitoring platform; analyzed vibration signals from accelerometer
sensors mounted on motors, pumps, fans, compressors, and gearboxes.
• Signal-to-image pipeline: cleaning, windowing, normalization, STFT spectrogram/scalogram generation, augmentation, validation dataset
preparation, and experiment tracking.
• Fault patterns: bearing wear, rotor imbalance, shaft misalignment, mechanical looseness, and gearbox anomalies appearing as frequency
peaks, harmonics, impulses, or non-stationary vibration energy.
• Models and outputs: trained ConvNeXt-style CNN and transformer-based experiments; prepared anomaly scores, predicted fault types,
confidence levels, severity estimates, and JSON outputs for dashboards/alerts.

Ophthalmology & Сlassification Image Analysis
Medical Image Analysis / Ophthalmology & Oncology Classification
Medical CV workflows for ophthalmology-related image classification and oncology-focused carcinoma / malignant calcification
suspicion detection in high-resolution medical images.
• Built preprocessing and training experiments for high-resolution medical images: ROI crops, intensity
normalization/windowing, augmentation policies, validation splits, class-imbalance handling, and model comparison reports.
• Adapted ConvNeXt/EfficientNet-style backbones for medical image classification tasks; experimented with label smoothing /
soft-label targets, PR-AUC / ROC-AUC monitoring, and threshold calibration for suspicious-case prioritization.
• Prepared product-ready outputs: pathology/suspicion probability, confidence score, suspicious-case prioritization, and API-
ready structured predictions for backend integration or analyst-review workflows.

Backend:
Reservation System / Backend Development
Commercial backend project for a recreation and resort complex with hotels, camping/tent areas, aqua-zone access, leisure
zones, and entertainment activities.
- Extended reservation functionality: tent/camping places, aqua-zone access, entertainment activities, and time-based
service slots with different availability rules.
- Implemented FastAPI backend logic: availability checks, booking lifecycle, status transitions, validation schemas, database
models, migrations, admin CRUD, and frontend/mobile API support.
- Worked with: Python, FastAPI, PostgreSQL, SQLAlchemy, Alembic, JWT authentication, and service/repository structure.

EDUCATION & LANGUAGES
B.Sc.-level studies, Faculty of Power and Aeronautical Engineering - Warsaw University of Technology | 2017 - 2021 | Warsaw,
Poland
Languages: Ukrainian - Native | English - Upper-Intermediate | Polish - Upper-Intermediate

Page 2