Nov 2025 — Present
Principal Software Engineer – New Lantern
- Lead the Viewer team (3 engineers) and own the imaging roadmap. Authored 575 merged PRs across frontend, backend, infra and ML services; reviewed 456 more from 15+ engineers.
- Launched HyperSync: cross-study registration on a Modal-hosted ANTs service (rigid + affine, scored by mutual information) that aligns priors to the current exam, so scrolling, measurements and trend charts follow anatomy. Cut registration from ~3 min to 16 s.
- Cut viewport interaction latency from 3,000 ms to 30 ms on 23,000-image studies and rebuilt frame caching from production telemetry. Platform p90 study load time fell 46% (19.4 s to 10.4 s).
- Rebuilt the admin experience as /manage, one in-product portal for users, studies, patients, sites, billing, audit logs, onboarding, report delivery and critical results. ~30 full-stack PRs in three weeks, plus the customer user guide.
- Built the phone app: worklist, chat, the viewer with touch controls, dictation and sign-off, priors, and on-call alerts within a minute of a study landing.
- Launched New Lantern on the Elgato Stream Deck Marketplace: 120+ actions on physical keys, 6 device profiles, live patient, RVU and chat dials.
- Led incident response and customer escalations across 9 customer groups, including commanding several SEV0 and SEV1 incidents.
- Built the viewer's observability stack (Datadog RUM, browser profiling, GPU telemetry, study-open timing) for a viewer that streams 16M+ image frames (7.7 TB) a week and opens several million studies a month.
- Expanded mammography CAD: Hologic EmphaSize, per-finding and case scores, one-click 2D-to-tomo navigation.
- Co-created the New Lantern Residency (239 completed applications) and led its application platform.
Oct 2024 — Nov 2025
Software Engineer – New Lantern
- Led the evolution of New Lantern's imaging viewer and product roadmap.
- Delivered a production-grade mammography suite with advanced tools, BI-RADS reporting, and radiologist-centric workflows, expanding New Lantern’s market and driving adoption.
- Designed a production-ready PET/CT and PET/MR fusion module with SUV quantification, tumor segmentation, and MIP 3D visualization, powering oncology workflows and extending New Lantern’s reach into molecular imaging.
- Pioneered a frontend performance overhaul that cut first-pixel load from 1.8s to under 100ms, with CDN edge warming and local caching bringing cached studies down to as low as 10 ms.
Show all 10 highlights
- Engineered a fully customizable and intelligent hanging protocol system, empowering radiologists to define, optimize, and personalize viewing layouts across all imaging modalities.
- Engineered high-performance multi-monitor support for the core radiology viewer, enabling seamless configuration and synchronized image display across up to four monitors.
- Integrated advanced AI models into clinical workflows, including SAM 2 for interactive medical image segmentation, TotalSegmentator for multi-organ labeling, and LLMs for structured report generation, accelerating diagnostic workflows and enabling intelligent automation.
- Led a large-scale refactor of the viewer’s architecture, introducing clean separation of concerns, modular builds, and scalable design patterns to dramatically reduce technical debt and improve maintainability.
- Served as the go-to expert for DICOM and medical imaging, contributing deep knowledge in image decoding, rendering optimization, and diagnostic workflow support across the platform.
- Established structured planning with estimates and milestones, improving engineering throughput.