Production systems I've engineered end-to-end — from autonomous AI agents and RAG pipelines to local LLM inference and OCR. Each card links to a deep-dive case study.
Mail-server gateway (Postfix Milter) that intercepts every inbound message, runs a fine-tuned BERT core (M-BSCE) through an asynchronous inference pipeline (RabbitMQ + Celery), and adds multimodal threat detection (PaddleOCR for Quishing) with SOAP clawbacks to the upstream mail system.
Autonomous multi-channel AI platform that unifies chat, voice, and email behind microservices and an API gateway, with shared context across channels.
Telecom recommendation agent fine-tuned on Qwen2.5-3B with LoRA, quantized to GGUF via Unsloth for cheap, low-latency inference.
LangGraph-powered agent that parses unstructured payroll emails, queries PostgreSQL, and routes edge cases to humans (HITL) for approval.
Voice-first enterprise assistant powered by vLLM-served LLMs, Pinecone vector search, Neo4j knowledge graph, and Whisper STT — answering HR and operations queries in real time.
Real-time avatar-based assistant combining LiveKit, WebRTC, Groq inference, and TTS/STT, with a Ready Player Me avatar for natural face-to-face interactions.
Chatbots that turn natural language into SQL, Pandas, and Neo4j Cypher queries against structured and relational data, with an Excel-AI-Chatbot on top of DuckDB.
Multi-agent evaluation framework that scores code changes by combining SonarQube quality metrics with Git activity signals, then surfaces the verdict to engineering managers.
End-to-end cheque digitization pipeline using OpenCV for image preprocessing, PaddleOCR for text and amount extraction, and a Flask + Docker orchestration layer.
Suite of context-aware document intelligence pipelines: PDF intelligence (PyMuPDF + spaCy + BERT), passport MRZ segmentation, and prescription OCR.