MLOps Lead · Zurich · since 2019
I build the path from training to deploy and rollback — not a manual ritual. Pipelines, registries, drift monitoring, and infrastructure as code. Currently lead MLOps at peoNova, Zug.
Selected
Contracts and public code. Filter by the kind of work, not the employer.
09 / 09
Hover a card to read the short description.
peoNova · Zug
2025 — present
AWS · GCP · Terraform · MLflow
In brief
Cloud stack for high-throughput AI apps: pipelines, a feature store, a model registry, and real-time drift monitoring. CI/CD with versioning cut model deployment time by 30%.
Public repository
Open source
FastAPI · Kubernetes · ONNX · Prometheus
In brief
Sentiment service with Helm, GitHub Actions, and observability. The production lesson: ship the model in the image, don’t pull it from S3 on every request.
RepositoryPublic repository
Open source
LangGraph · Milvus · MLflow · Ragas
In brief
Self-correcting agentic RAG: vector search, MLflow tracking, A/B tests, and Ragas quality gates inside CI — not after an incident.
RepositoryOpenZeppelin · Zug
2023 — 2024
Differential privacy · ZK · monitoring
In brief
Validation, testing, and monitoring before production. Data minimization and differential privacy cut PII exposure. Separate reviews of zero-knowledge systems for cryptographic weaknesses.
EncryptEdge Labs · Zurich
2023 — 2024
Deep learning · forecasting
In brief
A classification model cut error by 25% versus the legacy baseline. Demand forecasting: MAPE down 18%. Alongside that, customer segmentation and an automated training loop.
InnovateTech · Milan
2022
GAN · NLP · detection · recommendations
In brief
A text-to-image GAN, a recommender that lifted engagement 12%, and an NLP model that cut support response time by 33%. Object detection: mAP +25% in cluttered scenes.
Public repository
Open source
C++20 · artifact hashes
In brief
An append-only chain for model and dataset versions: cryptographic hashes and metadata on-chain, large weights off-chain in content-addressed storage.
RepositoryTechSolutions UA · Kyiv
2020 — 2021
Streaming · DWH · ETL · BI
In brief
Near-real-time event processing, anomaly alerts, and a cloud warehouse for a multinational client. ETL extraction 22% faster. A retention model flagged risk at 85% precision.
Public repository
Open source
Python · environment snapshots
In brief
Deterministic snapshots of a Python runtime, plus drift detection. Reproducibility lives in an artifact, not in whoever deployed last.
RepositoryAbout
Reproducible, observable, reversible — without those three, a model does not survive production.
Daniil Krizhanovskyi, mlops lead. Six years building production ML infrastructure across fintech, security, and enterprise data: Zurich, Zug, Rome, Milan, Kyiv.
Not a notebook demo — a system that holds load. Automated CI/CD, a feature store, a registry, metric drift, Terraform, and Kubernetes. Privacy is part of the platform, not a patch after the audit.
English is professional. Russian and Ukrainian are native. Italian is conversational.
2025 — present
Lead MLOps Engineer
peoNova · Zug
Contract
2023 — 2024
AI Safety & Solutions Engineer
OpenZeppelin · Zug
Concurrent contract
2023 — 2024
MLOps Engineer
EncryptEdge Labs · Zurich
Concurrent contract
2022 — 2023
MLOps Engineer
NextGen Technologies · Rome
Kubernetes and CI/CD
2022
ML Engineer
InnovateTech · Milan
Models in product
2020 — 2021
Data Engineer
TechSolutions UA · Kyiv
Streams, warehouse, ETL
2019 — 2020
Junior Data Analyst
TechSolutions UA · Kyiv
Dashboards and retention
2020 — 2022
Master’s, Applied Mathematics
Lviv Polytechnic National University
2020 — 2021
Master’s, Information Systems and Technologies
Odesa Polytechnic National University
2016 — 2019
Bachelor’s, Computer Engineering
Odesa Polytechnic National University
Skills
What I actually use to ship, watch, and move data. Certificates sit here so they aren’t buried at the end of a CV.
Python · Docker · Kubernetes · FastAPI · MLflow · SQL · Linux
CI/CD for ML · Terraform · Model registry · Drift detection · Observability · Experiment tracking
Streaming pipelines · Distributed systems · ETL · Feature store · REST APIs
PyTorch · ONNX · NLP · Forecasting · Detection · Recommendations
Contact
Send the task: a platform, a contract, or a review of what is already in production. The note does not leave this site — it opens in your mail client.