AI Automation Platform
Designed AI automation solutions for business clients — website chatbots, lead-gen agents, customer support automation, and workflow concepts. Authored technical and client-facing docs.
HelixWorksLLMsdocs
n8n Automation Workflows
Self-hosted n8n workflows integrating AI models, webhooks, WordPress, Slack, Telegram, Twilio, ElevenLabs, and custom APIs. Structured JSON outputs for multi-step business processes.
n8nwebhooksJSON
AI Content Generation Pipeline
Research → LLM generation → structured JSON → WordPress publishing → Slack editorial review. Scheduled generation, human approval, automated publishing.
LLMWordPressSlack
Local AI Infrastructure
Built and tested local inference using GPU hardware, Docker, Ollama, and OpenAI-compatible APIs. Integrated with automation workflows for private on-prem AI on sensitive data.
DockerOllamaGPU
Audio Transcription & Call Analysis
Local transcription + diarization pipeline using WhisperX, Pyannote, GPU acceleration, Docker, and n8n. Split transcription/analysis stages to cut model-load overhead.
WhisperXPyannoten8n
Legal AI Assistant / RAG
Privacy-focused legal assistant: OCR, document normalization, hybrid search, RAG indexing, local LLM inference. Chat-based querying of large case collections, data stays on-prem.
RAGOCRlocal LLM
Medical Voice Assistant (Concept)
Privacy-preserving voice assistant that locally transcribes doctor-patient conversations and drafts EMR notes on edge AI hardware, with mandatory human review.
edge AIEMRprivacy
Twilio · ElevenLabs · Telegram Voice Agent
Voice automation connecting Twilio, ElevenLabs voice agents, Telegram notifications, and webhook post-call processing. Answers calls, collects caller info, sends structured summaries.
TwilioElevenLabsTelegram
RAG & LightRAG Experiments
Built RAG-based retrieval workflows using LightRAG, custom Docker images, high-dimensional embeddings, and local/private AI infrastructure for document search.
LightRAGembeddingsDocker
Edge AI on Jetson Orin Nano
Tested edge deployment on NVIDIA Jetson Orin Nano — speech recognition models, CUDA/PyTorch setup, Dockerized inference services, FastAPI endpoints for local transcription.
JetsonCUDAFastAPI
Google Sheets Matching Automation
Sheets-based workflow matching origin addresses to optimal destinations — automated assignment by distance, travel efficiency, or business rules.
Sheetsroutingautomation
AI OCR & Doc-to-Markdown
OCR-to-Markdown workflows converting scanned docs into structured, machine-readable text — preserving layout, tables, checkboxes, headings for downstream AI / RAG ingestion.
OCRMarkdownRAG