AI Platform & MLOps Engineer
3+ years building and operating production ML, data and GenAI systems, with current work focused on agent reliability, evaluation, observability and platform control.
I take AI from a promising prototype to a system people rely on, inside the environment where it actually has to run.
Experience
MLOps & Analytics EngineerFibabanka, Analytics Center of ExcellenceMar 2023 — Mar 2026
- Owned deployment and lifecycle for 10 production ML models on Iguazio/MLRun, Docker and Kubernetes: validation, versioning, promotion, retraining, monitoring and containerized FastAPI serving.
- Re-architected sequential Oracle feature-engineering workflows into parallel dbt models orchestrated by Airflow, cutting end-to-end processing from 2–3 hours to 25–30 minutes.
- Built a centralized feature store and dimensional models across Oracle, Trino and SQL Server, with shared definitions, validation controls and data contracts reused by training, inference and analytics.
- Migrated validated Azure GenAI prototypes to on-premises GPU infrastructure, serving quantized open-source LLMs with llama.cpp and vLLM under data-residency constraints; contributed to 5 production LLM workflows.
- Built a GPU-distributed call-center pipeline: WhisperX transcription, timestamp alignment, speaker diarization and LLM summarization into structured KPI fields.
- Added Great Expectations and Deepchecks gates to Airflow pipelines and GitHub Actions delivery, so data and model regressions fail in the pipeline instead of surfacing in reports.
What I do
ML Platform & MLOps
Taking models from a notebook to production: training and scoring pipelines, model serving, versioned releases, monitoring and rollback on Kubernetes.
LLM, RAG & Agents
AI assistants and agents that work with company documents and tools, where code checks what the model produces before anything is used.
AI Reliability & Observability
Keeping AI systems dependable in production: evaluation on fixed test sets, a trace for every model call, release gates and safe failure.
Data Engineering
Batch and streaming pipelines with Kafka, dbt and SQL, with data quality checks and feature definitions shared across models.
Skills
- ML platform
- LLM & agents
- Data
- Observability
- Languages & APIs
Projects and what they are for
- ML PlatformAI Infrastructure / Platform EngineeringA self-hosted platform for running ML on Kubernetes: scheduled scoring jobs, versioned datasets, model monitoring and serving.
- Agentic Customer Service PlatformAI Reliability / Execution InfrastructureLets an AI agent handle refunds, cancellations and support tickets safely: the model suggests the action, code checks the rules and the customer confirms before anything runs.
- Knowledge Base RAGGenerative AI / RAG PlatformAnswers questions from long PDFs such as annual reports, 10-K filings and manuals, and shows the exact page each answer comes from.
- Agentic SREAgent Systems / AI PlatformInvestigates Kubernetes outages and points to the root cause with the evidence behind it, so on-call engineers do not have to dig through logs by hand.
- DecisionSQLStructured Data / Governed Text-to-SQLLets people ask data questions in plain English and get answers from read-only SQL that stays inside their access rules.
- ModelOps Control PlaneMLOps / AI PlatformRolls out new model versions gradually, checks their quality on live traffic and rolls them back automatically if they get worse.
- CauseTuneModel Training / Fine-TuningFine-tunes a small open-source language model for a specialist task and tests whether the improvement holds on cases it has never seen.
- Real-Time Commerce PlatformDistributed Systems / StreamingAn event-driven order system on Kafka that keeps orders and payments correct even when messages arrive twice or a service fails.
