Skip to content
View AmbujamSivan's full-sized avatar
💭
Working on Django Framework
💭
Working on Django Framework
0 followers · 1 following

Block or report AmbujamSivan

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AmbujamSivan/README.md

Hi, I'm Ambujam Sivan Pillai 👋

Full-stack engineer building agentic AI systems for hardware diagnostics and manufacturing. Based in Austin, TX.

I build LLM agents that do real work against real systems. They call tools, cross-check their own output, and fall back to deterministic logic when the model isn't sure. My focus is where AI meets the server management plane and the factory floor.

  • 🤖 Agentic AI: multi-agent orchestration with Semantic Kernel, MCP servers, RAG, LLM-as-judge grounding, function calling
  • 🖥️ Hardware and platform: Redfish/BMC, PCIe AER, POST/UEFI, and dmesg telemetry for automated root-cause analysis
  • 🏭 Manufacturing systems: SECS-II/GEM equipment integration and MES bridging
  • 🧱 Engineering: C#/.NET, Python, ASP.NET Core, gRPC, GraphQL, SignalR, Postgres/SQL Server, Docker, OpenTelemetry, xUnit

📌 Featured work

AgenticAI: Agentic AI monorepo

Project What it does
🔎 RCA Engine An agent that triages server hardware failures across five telemetry sources (Redfish, PCIe AER, dmesg, POST/UEFI, DPU console) and produces a structured RCA report. It has a deterministic fallback, so a report always ships.
🏭 AeroMind IQ Five Semantic Kernel agents that investigate production anomalies. An Isolation Forest flags the anomalies, a critic reviews SQL before it runs, and an LLM-as-judge checks the report for groundedness. Traced with OpenTelemetry and Langfuse.
🩺 Redfish Diagnostic Emulator A mock BMC that follows the DMTF Redfish spec, with fault injection. It's covered by unit, contract-conformance, and stress test suites.
🔌 Communication Protocols Lab One domain served over REST, GraphQL, gRPC, WebSocket, and SignalR. On top of that: an MCP server, a RAG layer on Qdrant, and a Semantic Kernel agent.

FabBridgeEngine: SECS-II → MES bridge

A C#/.NET 8 service that turns equipment collection events (S6F11/CEID) into MES operational states and persists them to SQL Server. Live overview →

🛠️ How I build

  • Evidence over vibes. Agents ground their claims in tool output, and a separate judge grades them.
  • Always ship a result. Every LLM path has a deterministic fallback.
  • Provider-agnostic. Switch between Ollama (local), OpenAI, Azure OpenAI, Gemini, and Claude in config.
  • Observable by default. Traces, token counts, and cost guardrails are built in.

Built with AI-assisted development (Claude, Copilot), in line with modern engineering practice.

📫 Let's connect: LinkedIn

Pinned Loading

  1. AgenticAI AgenticAI Public

    Agentic AI monorepo: multi-agent Semantic Kernel systems, MCP + RAG, LLM tool-calling for server hardware diagnostics (Redfish BMC, PCIe AER, RCA triage) in C#/.NET and Python.

    C# 1

  2. FabBridgeEngine FabBridgeEngine Public

    SECS-II → MES bridge in C#/.NET 8: ingests equipment collection events (S6F11/CEID), maps them to MES operational states, and persists to SQL Server.

    C# 1