Ali D.

Most teams don't need more AI ideas. They need AI that actually works in production, integrated into their existing tools, automated end-to-end, and built to survive real data and real users. I'm Ali Dabouei (PhD), a full-stack AI engineer based in Vienna, VA. I own the entire delivery: model design, pipeline architecture, backend/frontend development, API integrations, automation workflows, and cloud or on-prem deployment. One person, full accountability, prototype to production. What I build: Agentic AI · RAG pipelines · Voice AI agents · Computer vision · OCR & Document AI · Generative AI · AI automation & API integrations Recent work: • Logistics OCR platform: 10,000+ invoices/week at 98%+ accuracy, with automated auditing and billing discrepancy detection • Agentic legal framework: native legal CRM integration automating 60%+ of routine legal tasks, no Zapier or n8n • Voice AI agents: deployed across healthcare scheduling, B2B sales qualification, real estate lead follow-up, and 24/7 restaurant order handling, all with full CRM, calendar, and operations dashboard integration. • RAG assistants: on-prem document analysis, e-commerce support across chat/email/social, and adaptive tutoring • Sports video analytics: YOLO + SAM pipeline extracting player tracking and tactical metrics from raw game footage Tech stack: Python · FastAPI · LangGraph/LangChain · React · Next.js · PostgreSQL · AWS · Twilio · ElevenLabs · Deepgram · Pinecone · Weaviate · PaddleOCR · PyTorch · OpenCV · YOLO · Amazon Bedrock · Google Gemini · REST APIs My background is research-heavy (PhD WVU, postdoc CMU, 30+ publications) but my style is practical: fast prototype → tight feedback loop → rigorous stress-testing before launch. I build automations natively, integrate directly with your APIs and CRMs, and have on-prem deployment experience for security-sensitive environments (legal, healthcare, finance). Typically, I start with a short discovery call or scoped paid milestone to validate the approach before committing to a full build.