Michael Hafen

Senior AI/ML Engineer

AlphaSense

Sep 2024 - Jan 2026 (1 year 5 months)

• Championed end-to-end delivery of developer-facing Gradient Platform features using TypeScript, React, and GraphQL, shipping 6 core UI components and reducing developer integration time by 42%. • Spearheaded maintenance and performance improvements for Go microservices and GraphQL API, increasing throughput by 35% and lowering p95 latency by 28% across production traffic. • Architected Python model-serving endpoints with FastAPI and gRPC, enabling 24/7 inference at 200+ requests per second and supporting rollbackable model deployments. • Authored a TypeScript design system and Storybook-driven component library, increasing UI reuse across 12 projects and cutting visual defects by 47%. • Coordinated cross-functional design and UX reviews f

Deep Learning Engineer

Tempus AI

Aug 2023 - Jun 2024 (11 months)

• Designed and optimized distributed training pipelines in Python using PyTorch DDP, cutting full-epoch training time by 67% on 64-GPU clusters. • Built React and TypeScript tooling to visualize model explanations and dataset distributions, adopted by 18 clinicians weekly and improving diagnostic review speed by 31%. • Optimized ETL into Snowflake and PostgreSQL for imaging metadata, reducing ingestion latency by 73% while supporting 2 TB of daily data. • Deployed models with Docker, Kubernetes, and MLflow, supporting continuous deployment of 14 model versions with automated rollback and canary strategies. • Collaborated with UX and product teams to design research dashboards and standardized components, improving user task completion by 29

Data Scientist

Veritone

May 2019 - Mar 2023 (3 years 11 months)

• Led architecture and delivery of scalable model-serving platforms in Python, supporting 150+ models and handling 10,000+ daily inference requests with autoscaling. • Engineered REST and GraphQL APIs in Go and Python to expose ML capabilities, increasing enterprise feature adoption by 62% across customers. • Developed ETL pipelines with Airflow and Kafka to process 50 million media events per month, decreasing batch processing latency by 55%. • Built data warehouses and analytics pipelines in Redshift and Snowflake to support experimentation and reporting, reducing analytical query time by 70%. • Orchestrated containerization and runtime strategy using Docker and Kubernetes, improving node utilization by 38% and lowering per-service memory

F

ML Engineer

Flexiv

Jun 2017 - Feb 2019 (1 year 9 months)

• Designed real-time control models and deployed Python inference services on embedded Linux, improving robotic task accuracy by 21%. • Implemented CI/CD for model releases using Jenkins and Docker, cutting release timelines from weeks to 48 hours for edge deployments. • Integrated telemetry and metrics pipelines using InfluxDB and Grafana to monitor model drift across 30 robots, reducing downtime by 39%. • Developed data capture and labeling pipelines with MySQL and Amazon S3, building a 600k-sample dataset to accelerate supervised training iterations. • Refactored legacy codebase to Python 3, added type hints and static analysis with mypy, raising static coverage to 87% and lowering runtime errors. • Supported agile product sprints to ali

Junior ML Engineer

HubSpot

Nov 2016 - Mar 2017 (5 months)

• Assisted in developing recommendation models in Python, increasing click-through rates by 12% during pilot deployments. • Supported REST API integrations for model scoring into internal tools using Flask, reducing integration time by 66% for product teams. • Created preprocessing pipelines with Pandas and PostgreSQL to normalize datasets for A/B testing across 8 product teams. • Collaborated closely with engineers to design feature stores and selection strategies, improving model training throughput by 31%. • Prototyped visualization widgets in React and plain JavaScript to help product managers interpret model outputs during demos. • Benchmark-tested model variants and reported clear metrics, contributing to a 9% improvement in baseline

About

Detail

Timeline

Résumé