|
Enterprise email security with LLM-powered mail triage
|
Healthcare anomaly detection & ML pipelines
|
๐ก๏ธ MailShieldAI Highlights
- ๐ง Crafted modular LangChain adapters for MCP tool ecosystems and message pipelines, slashing legacy code by >70%
- ๐ Built streaming-safe message translators between Vercel AI SDK clients and LangChain backends
- ๐ Delivered Jinja-backed prompt systems with multi-model support (OpenAI-compatible + Ollama)
- ๐งช Extensive pytest coverage and UV-managed builds for reliability
- ๐ Shipped TypeScript/Next.js frontends, Go microservices, and Python-based synthetic evaluators
๐ฅ BinaryMedAI Highlights
- ๐๏ธ Reimagined orchestration layer: FastAPI routers โ service logic โ MongoDB/ADM/PDFM adapters
- ๐ Added on-demand PDF generation, schedule lifecycle management, and DEBUG-mode mock data
- ๐ณ Standardized deployment with Docker Compose, UV package management, and network-aware
.envbootstraps - ๐ Authored expandable documentation so the medical ops team can self-serve
const currentFocus = {
security: "Agentic workflows mixing static detection with human-in-the-loop review",
redTeaming: "LLM-powered testing for phishing and medical anomaly pipelines",
compliance: "Safer deployment playbooks for regulated environments (SOC2, HIPAA)"
};I'm always up for pairing on security tooling, ML evaluation frameworks, or developer experience upgrades.



