Our Portfolio

Implemented AI agents to automate DOT compliance monitoring, audit preparation, and driver file management. The system processes thousands of documents daily and flags potential issues before they become violations.

Built an AI-powered SOC 2 compliance platform that automates evidence collection, policy generation, and audit trail management. The system continuously monitors compliance posture and generates readiness reports.

Deployed intelligent AI agents that handle customer inquiries across email, chat, and phone. The agents understand context, access customer history, and resolve complex issues without human intervention.

Built an edge-deployed computer vision system that identifies cafeteria tray contents at the point of sale. YOLO object-detection models running on NVIDIA Jetson hardware classify 80-100 food and product classes in real time, integrated with POS tills via a FastAPI service to automate checkout, per-till menu mapping, and live analytics.

Built a fully operational AI-powered SAP (Substance Abuse Professional) evaluation assistant that automates session recording (Zoom), transcription, ASI-6 form completion, and compliance documentation with DocuSign signing and Box secure storage for DOT return-to-duty evaluations. A Node.js/Express backend with PostgreSQL pairs with a React 19 frontend, orchestrated by a multi-agent AI system running on a local Llama 3.2 model via Ollama — five specialized agents handle transcription, sentiment analysis, risk detection, report generation, and question guidance. Socket.IO WebSocket powers live transcript streaming, session presence, and instant notifications, all secured with JWT auth, audit logging, and rate limiting for HIPAA-compliant operations.

Built a multi-service AI-native platform for a tri-state restoration company spanning a static Next.js marketing site generating 300+ SEO pages from Sanity, an AI phone agent handling after-hours calls across 13 numbers in three area codes, a GPU-powered content generation pipeline, and an AWS serverless intelligence backend. The Node.js/Twilio phone agent qualifies leads, detects emergencies via keyword + sentiment analysis, and escalates critical calls with live transfer and SMS. A Python/vLLM content engine generated and QC'd 207 SEO-optimized pages in a single hands-off batch run with automated readability, keyword-density, and similarity gates.