Senior AI/ML Engineer at Cisco

Architected high-availability AI infrastructure in a regulated enterprise with multi-business unit demands by implementing load balancing, autoscaling, and circuit-breakers, achieving 99.9% uptime SLAs and reducing tail-latency for production deployments by 40%. Engineered end-to-end model governance workflows using MLflow registry and immutable artifacts for validation, versioning, and rollback, cutting deployment errors and streamlining audits by 40%. Built shared feature platforms with Feast and custom orchestration for reusable features across teams, improving training/serving efficiency and reducing feature inconsistencies by 50%. Deployed model monitoring with Prometheus/Grafana to detect drift and trigger alerts, minimizing false negatives and boosting overall model accuracy by 30%. Collaborated with security, infrastructure, and legal to embed access controls and encryption in AI pipelines to meet regulatory requirements, enhancing governance and reducing compliance risks.

Applied AI Engineer Front at Front

Created AI-augmented workflows to alleviate inbox overload for 10k+ daily users by introducing prioritization, tagging, and scoring into product paths, increasing user engagement with personalized features and reducing processing delays. Optimized scalable ML inference APIs for sub-200ms latency under noisy inputs using TF Serving, FastAPI, caching, and fallbacks, reducing latency by 40% and improving reliability. Transformed raw event logs into versioned training datasets enabling A/B testing and controlled rollouts, accelerating model updates and improving prediction accuracy by 25%.

Data Engineer, ML Platform at Catch&Release

Orchestrated ingestion pipelines across heterogeneous sources, incorporating cleansing and schema evolution to support AI workloads, cutting data processing time by 50%. Established dataset versioning, quality checks, and backfills to ensure reliable model training amid growth, decreasing pipeline failures by 30% and improving downstream model quality. Implemented feature infrastructure serving product teams with Luigi-inspired patterns, enabling production inference APIs and boosting feature reuse while reducing development cycles by 40%. Monitored observability with metrics, logs, and alerts to tackle anomaly risks in fast-paced ops by detecting failures early, enhancing platform reliability and reducing downtime by 25%.