Senior Data / AI Platform Engineer at Humanloop

Architected and owned backend data services for AI prompts, evaluation runs, feedback records, and model traces; redesigned async processing for outputs and production usage events to reduce latency; implemented vector search workflows with embeddings, Postgres, Redis; hardened workspace controls with RBAC, audit logging, encrypted secrets, tenant isolation, and GDPR‑aware data handling; partnered with product and research teams to turn evaluation workflows into stable APIs, dashboards, and data models; mentored engineers on Python services, data modeling, testing strategy, and production debugging.

Senior Data / ML Platform Engineer at Quantexa

Architected and owned backend data services for AI prompt version evaluation runs, feedback records, and model traces; built Spark pipelines for entity resolution, transaction networks, customer relationships, and investigation datasets; developed Scala and Python jobs for ingestion, matching, enrichment, and analytic feature generation; improved fraud detection accuracy by 15%; tuned Spark jobs, partitions, and cluster resources; introduced Kafka-backed ingestion for near real-time data; standardized data quality checks for schema drift, duplicates, missing identifiers, and inconsistent attributes; supported ML feature pipelines for risk scoring, anomaly detection, and prioritized investigations; reduced incident response time by 25% and mentored data engineers on Python services, data modeling, testing strategy, and production debugging.

Data Platform Engineer at Sky

Scaled analytics pipelines for viewing behavior, customer events, subscriptions, and content engagement (100TB+ daily); built PySpark and Airflow workflows for batch and streaming data from product and customer systems, ensuring data freshness and accuracy; produced curated datasets for analytics, marketing, retention, and product performance reporting; migrated legacy ETL jobs to Airflow DAGs with clear ownership, retries, and failure alerts, improving operational reliability.