I build production‑ready back‑end services and MLOps pipelines that turn research‑grade models into business value. My background in AI engineering (MSc, Ural Federal University) helps me speak both the language of data science and of scalable software. I love clean architecture, well‑documented APIs and CI/CD that “just works”.
Currently looking for a team where engineering rigour and curiosity meet to deliver products that matter.
- 🚀 Delivered projects – from proof‑of‑concept to cloud deployment.
- 🔄 Full‑cycle ownership – requirements → system design → coding → tests → DevOps.
- ⚡ Process optimisation – automated data & model pipelines saved analysts 30 + hrs/month.
- 🤝 Agile & teamwork – experienced in Scrum/Kanban, peer reviews and trunk‑based Git flow.
| 🗂️ Area | ⚙️ Stack |
|---|---|
| Languages | Python (primary) · C# |
| Web / API | Django & DRF · FastAPI · Flask |
| Data & AI | PyTorch · scikit‑learn · HuggingFace · DeepLabCut |
| MLOps | DVC · Prefect · MLflow · GitHub Actions · Jenkins · GitLab CI |
| Databases & Queues | PostgreSQL · Redis · RabbitMQ · Celery |
| DevOps | Docker & Compose · Linux · Git · Git LFS |
| Project | What it does | Business impact | Key tech |
|---|---|---|---|
Video Upscale & Frame‑Interpolation Platformdjango‑upscale‑interpolate‑videos |
Django + Celery orchestrates RIFE & ESRGAN micro‑services for video quality enhancement. | Cut manual post‑processing time at a media studio by 40 %. | Django DRF · Celery · RIFE · ESRGAN · PostgreSQL · Docker‑Compose · GitHub Actions |
| Garage Automation Suite (private) |
End‑to‑end automation for a car‑service chain: live queue, booking, services catalogue, BERT‑based symptom assistant. | Shortened reception time per client by 30 % and improved utilisation forecasts. | Django · BERT fine‑tuning · REST API · Docker |
| Project | What it does | Challenge solved | Key tech |
|---|---|---|---|
Newts Labellingnewts_labelling |
Prefect flow downloads, aggregates and resizes wildlife images; initialises a DeepLabCut project; syncs with Google Drive. | Replaced a brittle manual workflow – 5× faster dataset refresh. | Prefect · DeepLabCut · Google Drive API · Docker |
DeepLabCut Inference & Croppingdeeplabcut_images_inference_and_cropping |
Streamlit app resizes images, runs DLC inference, crops regions (e.g., newt bellies). | Batch GPU inference within memory limits; automated ROI extraction. | Streamlit · DeepLabCut · OpenCV |
Streamlit Multi‑Label Classifierstreamlit_multi-label_classifier |
EfficientNet‑B4 model with visual dashboard for multi‑tagging marketing images. | Enabled non‑tech staff to tag assets with > 95 % precision. | Streamlit · PyTorch · Altair |
| Project | What it does | Notes | Key tech |
|---|---|---|---|
Titanic Survival Prediction APIMLOps_team21_final |
FastAPI service predicts survival from passenger data; CI trains & deploys model with DVC‑versioned data. | University capstone demonstrating end‑to‑end MLOps best practices. | FastAPI · scikit‑learn · DVC · Jenkins · Docker |
- Rogue‑like Prototype – inventory, stat system & item interaction.
- Simple Coffee Visual Novel – gameplay & state management (dialogue engine).
- Start with the why – clarify business goals before writing code.
- Design first – draw sequence & deployment diagrams; pick the simplest architecture that scales.
- Automate everything – tests, linting, data & model versioning, container builds.
- Measure & iterate – ship small, collect metrics, improve.
- ✉️ Email: kitathedeveloper@gmail.com
- 💬 Telegram: @restfulpanda
- 🔗 LinkedIn: on request
Thanks for reading – looking forward to building great things together! 🚀