Hi, I'm Sean Pesis 👋

AI Software Engineer — building production LLM systems

🚀 Live Project: CareerCoach AI • 🌐 Portfolio • 💼 LinkedIn • ✉️ Email

---

🧠 About Me

I build full-stack systems that put LLMs into real production workflows — not demos, not notebooks. My day job is engineering internal platforms at Herzliya Medical Center, where I ship C#/.NET and React systems that integrate Claude and MCP servers to automate document and operations workflows for clinical staff. On the side, I build AI-first web products (see CareerCoach AI below).

• 🤖 AI Focus: Production RAG pipelines, vector search, LLM orchestration (Claude, OpenAI), prompt engineering with real evals.
• 🏗️ Backend: C#/.NET, Node.js, Next.js — REST and WebSocket services, microservices, real-time systems.
• 🏥 Domain: Five years inside healthcare IT — comfortable with reliability, security, and the messy reality of shipping to non-technical users.
• 🎓 Education: B.Sc. Computer Science, Holon Institute of Technology (HIT) — expected March 2026.

---

🚀 Featured Project — CareerCoach AI

A production SaaS that helps job seekers improve their resumes with AI feedback grounded in their actual document.

Stack Frontend: Next.js 16, TypeScript, Tailwind AI: Claude 3.5 Sonnet, custom RAG pipeline Vector DB: Pinecone (semantic retrieval over uploaded PDFs) Auth & Payments: Clerk, Stripe Infra: Vercel

What it actually does Ingests a user's resume PDF and embeds it for retrieval RAG layer grounds every Claude response in the user's own document — preventing hallucinated advice Streams structured feedback (clarity, ATS-fit, weak bullets) back to the UI Paywall + auth flow handled end-to-end

Why it matters: most "AI resume tools" generate generic advice. The RAG grounding is what makes the feedback actually about your resume — and shipping that pipeline reliably is the interesting engineering problem.

🔗 Try it live →

---

🏥 Production Work at Herzliya Medical Center

Repos are private (hospital infrastructure), but the work is real:

• MdSignage — C# / .NET 8, ASP.NET Core, Blazor, PostgreSQL AI-enhanced digital signage platform controlling 18 BrightSign players across the hospital, replacing a third-party SaaS dependency. LLM-powered ingestion pipeline parses Excel and scanned PDF shift rosters into structured schedule data.

• MD Scan — C# / .NET Framework 4.8, TWAIN protocol, WebSocket/WSS Browser-integrated document scanning system using the TWAIN protocol over a secure WebSocket bridge, with LLM-based post-processing on scanned documents.

---

🛠️ Tech Stack

🤖 AI & LLMs

_{RAG pipelines · Pinecone · Claude API · OpenAI · prompt engineering · LLM streaming · embeddings · semantic search · Anthropic MCP servers}

💻 Full-Stack

_{C# · .NET / ASP.NET Core · Blazor · Next.js · React · TypeScript · Node.js · REST · WebSockets}

🗄️ Data

☁️ Infra & DevOps

---

📊 GitHub Stats

---

🤝 Open to Collaborate On

• AI-first products — especially anything with non-trivial RAG, retrieval quality, or LLM evals
• Healthcare / regulated-domain tooling where reliability and UX both matter
• C#/.NET ↔ Python AI integrations — bridging enterprise backends to modern AI stacks

_{Tel Aviv, IL · Last updated May 2026}