Founding Engineer and de facto Co-Founder with 7+ years of software engineering experience. As the first technical hire and sole engineer for 1.5 years, built the entire AI diagnostic platform from concept to production, leading all technical architecture, ML development, and infrastructure decisions. Successfully delivered a production-grade healthcare AI system serving dermatology clinics, demonstrating full-stack capabilities from data engineering to cloud architecture and ML operations. Proven ability to single-handedly architect complex enterprise systems while navigating healthcare regulations and securing EHR partnerships.
- Architected a serverless, event-driven ML inference system using AWS Lambda, DynamoDB, EventBridge, and Step Functions
- Designed a complex multi-step polling mechanism to overcome EHR API limitations (no webhook support)
- Implemented Infrastructure as Code (IaC) using AWS CDK for reproducible, scalable deployments
- Built real-time monitoring system processing clinic encounters every few seconds with sub-second latency
- Achieved 99.9% uptime in production environment serving dermatology clinics
- Successfully integrated with Modernizing Medicine (ModMed) EHR platform, serving 40%+ of US dermatology market
- Navigated complex healthcare compliance requirements including HIPAA and Business Associate Agreements
- Implemented secure bidirectional data flow between cloud infrastructure and clinical systems
- Achieved third-party vendor approval through rigorous sandbox testing and architecture review
- Processed millions of patient records, visits, and medical images from production EHR systems
- Built highly parallel crawler system to overcome API throttling and extract maximum data
- Designed data lake architecture handling pathology images with metadata
- Implemented efficient data pipelines processing 3M+ additional pathology results
- Created custom database migrations and ETL pipelines for FHIR-compliant healthcare data
- Developed automated data quality gates and validation pipelines
- Implemented comprehensive data governance framework for healthcare data
- Built custom image quality assessment system with 98% true positive rate for clinical relevance
- Established data versioning and lineage tracking for ML reproducibility
- Built end-to-end computer vision pipeline for dermatological image analysis
- Trained custom binary classification model achieving 98% TPR for clinical relevance detection
- Developed object detection model (mAP@50: 55%) for automatic lesion localization and cropping
- Implemented transfer learning approaches using ResNet, MobileNet, and SOTA foundation models
- Achieved 70-80% weighted accuracy on 65-class dermatological condition classification
- Applied data clustering and dimensionality reduction for class taxonomy optimization
- Implemented sophisticated data augmentation and preprocessing pipelines
- Developed ensemble methods combining multiple models for improved diagnostic accuracy
- Utilized foundation models and transfer learning to overcome limited training data
- Built custom evaluation frameworks with comprehensive metrics and logging
- Designed and deployed full-stack web application for dermatologist image annotation
- Built client-side React application with real-time image viewing and labeling
- Developed Node.js backend with PostgreSQL for efficient data management
- Implemented LLM-powered validation system for accelerating annotation workflow
- Processed expert-annotated images for model improvement
- Created comprehensive ML experiment tracking with Weights & Biases integration
- Built custom CLI tools and scripts for data processing and model evaluation
- Implemented automated testing and CI/CD pipelines for ML systems
- Developed monitoring and alerting systems for production ML models
- Founded and built entire technical platform as first engineer, establishing all technical decisions and architecture
- Served as sole technical lead for all engineering functions during first 1.5 years, making every architectural and implementation decision
- Scaled technical leadership to manage and mentor team of 3-4 engineers (AI Engineer, Software Engineer, Data Engineer) as company grew
- Single-handedly developed MVP from concept to production while managing all technical stakeholder relationships
- Established company's technical foundation enabling Series A readiness and commercial scaling
- Built and led engineering culture, hiring practices, and technical roadmap as team expanded
- Identified and solved critical data access limitations by building custom API crawlers
- Overcame EHR API constraints through innovative polling and queue-based architecture
- Resolved data quality issues through multi-stage filtering and validation systems
- Addressed class imbalance through strategic data curation and taxonomy redesign
- Turned 40-50% model accuracy into 70-80% through systematic improvements
- Led technical strategy for MVP development and production readiness
- Made key architectural decisions balancing scalability, cost, and maintainability
- Innovated novel approaches to healthcare data integration and ML deployment
- Established best practices for MLOps and reproducible ML in regulated environments
- Mastered healthcare data standards including FHIR and medical coding systems
- Processed and understood complex pathology reports and clinical documentation
- Worked directly with dermatologists to understand clinical requirements and constraints
- Translated medical expertise into technical requirements and ML model design
- Ensured HIPAA compliance throughout data processing and model deployment
- Navigated complex healthcare vendor approval processes
- Implemented audit trails and security measures for protected health information
- Established data governance policies meeting healthcare industry standards
- AWS: Lambda, DynamoDB, S3, EventBridge, Step Functions, CDK, CloudWatch
- Infrastructure as Code: AWS CDK, Terraform concepts
- Monitoring: CloudWatch, custom logging solutions
- Languages: Python, JavaScript/Node.js, SQL
- Databases: PostgreSQL, DynamoDB, data lake architectures
- APIs: REST, GraphQL, healthcare-specific APIs (FHIR)
- Data Processing: Pandas, NumPy, Apache Spark concepts
- Frameworks: PyTorch, TensorFlow, scikit-learn
- Computer Vision: OpenCV, custom CNN architectures, transfer learning
- MLOps: Weights & Biases, MLflow concepts, Docker
- Foundation Models: SOTA model integration and fine-tuning
- Frontend: React, modern web technologies
- Version Control: Git, GitHub
- CI/CD: GitHub Actions, automated testing
- Documentation: Comprehensive technical documentation
- Scaled from prototype to production handling millions of patient records
- Improved model accuracy from 40-50% to 70-80% through systematic optimization
- Built systems processing 3M+ pathology results with 98% quality filtering accuracy
- Reduced manual data processing time by through automation
- Currently in active production serving dermatology clinics with live patient data
- Founded technical foundation for AI healthcare startup from ground zero
- Attracted major EHR partnership (ModMed) through technical excellence and vision
- Established technical foundation enabling funding-round readiness and commercial expansion
- Successfully deployed system to production clinics with active real-world usage
- Invented polling-based EHR integration architecture overcoming API limitations
- Developed multi-stage image quality filtering system achieving 98% TPR
- Created hybrid human-AML annotation workflow accelerating data labeling 10x
- Designed lesion localization system improving diagnostic accuracy 30%
- Built scalable ML inference platform for healthcare applications
- Founding Engineering: Built entire technical platform as first engineer and scaled to manage 3-4 person team
- System Architecture: Production ML systems with serverless AWS infrastructure at enterprise scale
- Computer Vision: End-to-end pipeline from medical image processing to diagnostic classification
- Healthcare Domain: Deep expertise with EHR integration, HIPAA compliance, and medical data systems
- Technical Leadership: Solo founder to team lead with hiring, culture, and technical roadmap ownership
- Full-Stack Capability: Data engineering to cloud infrastructure to clinician-facing applications