Senior Full-Stack Engineer | AI & Data Systems

I build AI automation systems on a part-time basis. I developed a fully automated, multi-agent sales pipeline from initial contact through offer generation using LangGraph and n8n for orchestration, Pinecone for retrieval, Instantly.ai for outreach delivery, and Monday.com for status tracking.

I also built reusable Airtable and Twilio automations for client outreach and lead workflows. I kept business rules configurable and separate from application code to support operational flexibility.

I served as technical lead across five repositories for a skilled nursing healthcare monitoring platform, including Flutter tablet applications, a React administration portal, a Laravel billing and integration portal, and Node.js services. I led development, performed code reviews, and mentored junior developers.

I built CPM, an internal radar-based resident monitoring system with custom TCP ingestion at 20Hz, pandas-based feature extraction, and an eight-stage SQL alert pipeline that compared each resident to their own baseline. I also designed a four-tier hot, warm, cold, and archive data-retention architecture.

I integrated PointClickCare using OAuth2 PKCE and processed 14 FHIR R4 ADT event types for admissions and clinical-record synchronization. I built an AI-assisted clinical documentation layer for RPM billing notes, progress notes, and care plans using Azure OpenAI, pgvector, LangChain, and a LangGraph verification workflow with mandatory human sign-off.

I led AI development for a rescue-operations platform through Xtensa. I developed a FastAPI backend and implemented PyTorch and Hugging Face model-training workflows, Whisper-based speech processing, and CUDA-accelerated training loops.

I improved OCR extraction accuracy from 70% to 92% and migrated the retrieval layer to Pinecone. I also built an iOS mobile client in Swift using AVAudioEngine for on-device audio capture and voice analysis.

I worked on a live B2B automotive e-commerce platform through KNZ, supporting Magento 2.4.7 with more than 100 million product images and over 10 million transactions. I diagnosed production 504 outages caused by full-table scans and resolved them through indexing, query rewrites, and caching, reducing page loads from more than 10 seconds to under 200 milliseconds.

I built a dedicated AWS data platform using RDS, S3, ECS, and Fargate to move reporting workloads off the production database. The platform included Python and pandas ETL, a Snowflake warehouse with RAW, STAGING, CORE, and MART layers, dbt transformations, and Airflow/MWAA orchestration.

I implemented incremental and idempotent loading using watermarks and Snowflake MERGE operations on stable business keys. I created SCD Type 2 dimensional models, including a vehicle-to-part fitment bridge table, and managed infrastructure with Terraform and custom NGINX configuration.