AI/ML Engineer &
Backend Developer

At Cedara, I develop for a carbon intelligence platform in the advertising vertical, helping companies achieve NetZero targets. My role bridges high-scale backend development with AI/ML infrastructure.

The core challenge was transitioning an advertising campaign optimization solution from a Python/PySpark proof-of-concept into a production-grade system. I rebuilt the pipeline in Scala on SparkML, deployed on Google Cloud Dataproc, delivering reliable and scalable ML inference at production volume.

Beyond the ML work, I architected Laravel-based batch processes for large-scale data ingestion and built reactive client-facing interfaces with Livewire. A key contribution was developing logic that uses carbon footprint data as a proxy for media value — a central piece of the platform's optimization engine.

From rapid prototyping to production-grade deployment on cloud infrastructure.

At Bandai Namco Entertainment Romania, I built a RESTful API from scratch for the mobile game Retro Wonder Park. The API served as the entire backend layer enabling gameplay features, player state management, and server-side game logic.

The project demanded a clean, well-documented contract between the mobile client and backend from day one. I designed the full resource model, endpoint structure, and request/response lifecycle, backed by comprehensive functional testing to ensure reliability across every game flow.

Starting from zero meant making foundational architectural decisions — authentication schemes, data persistence strategy, error handling conventions — decisions that would be lived with for the product's entire lifetime.

Clean architecture from the ground up, with testing built into every layer.

As a senior consultant at Pentalog, I was embedded in a leading European financial institution. I developed and maintained features within a large-scale Service-Oriented Architecture handling real financial transactions.

I championed a rigorous quality culture through Test-Driven Development, maintaining 99% unit test coverage across my services. This discipline paid off: our team achieved bi-weekly deployments for six consecutive months with a 0% error rate — a remarkable result in a regulated fintech environment.

Beyond writing code, I implemented Datadog monitoring for application observability and contributed to architectural decisions across the SOA. Working within a regulated platform sharpened my focus on reliability, traceability, and defensive coding practices.

Embed, elevate quality, and deliver with zero-defect discipline.

At eMAG, Romania's largest e-commerce platform, I was part of the Search Team maintaining Apache Solr search infrastructure. A major initiative was rewriting the Search API as an independent microservice, decoupling it from the monolith and integrating asynchronously with documenting, targeting, and anti-spy services.

I architected a parallel regression testing pipeline using Docker and Selenium Grid that reduced the test suite execution time by 90% — a direct, measurable performance improvement that accelerated the team's delivery cycle.

At eMAG's Future Hacks 2.0 hackathon, I won the Innovation Award (AI/ML) by building a proof-of-concept that predicted user funnel stages and conversions from web access logs using R (logistic regression) and Python (Random Forests).

Measure first, optimize precisely, and prove the impact with numbers.