Hire Justin F. - AI Engineer - New York

Designed and developed backend services for Deep Manifold’s AI platform, enabling large-scale data ingestion, feature computation, and model deployment for buy-side analytics. Built RESTful APIs and microservices using Python (FastAPI) and GCP Cloud Run, improving scalability and system reliability. Integrated machine learning pipelines for model training, evaluation, and serving within an automated MLOps framework. Collaborated with data scientists to productionize research models, optimizing latency and performance for real-time financial data processing. Contributed to LLM-based consulting tools, including retrieval-augmented generation (RAG) and domain-specific Copilot solutions.

Designed and implemented large-scale distributed data processing pipelines on Apache Spark and Delta Lake, improving query latency by 30% across internal analytics workloads. Built RESTful microservices in Scala and Python (Flask) to automate job orchestration and metadata management for ML experiments on MLflow. Collaborated with platform and infra teams to containerize and deploy Spark clusters using Kubernetes and Databricks Runtime, ensuring reliability under heavy concurrent workloads. Developed monitoring and alerting dashboards using Prometheus and Grafana, reducing on-call response time by 25%. Contributed to an internal feature-store API enabling unified feature access for model training and online inference, used by 10+ ML teams.

Designed and implemented scalable backend services in Go and Python, powering real-time content delivery and user interaction features for Snapchat Discover and Stories. Optimized recommendation pipelines using Spark and Airflow, reducing average feed generation latency by 35% while improving engagement metrics across key user cohorts. Built and maintained event-driven data pipelines on Kafka and GCP BigQuery, supporting petabyte-scale analytics and personalized ad targeting. Collaborated with ML and product teams to launch interest-based content ranking models, integrating feature extraction services and online inference APIs. Enhanced system observability and fault tolerance through Datadog dashboards, structured logging, and automated incident recovery, cutting downtime by 40%.