I am a pre-sales Solution Engineer working on agentic Ai workflows and multi-agent systems that solve real business problems
I aim at bridging the gap between complex Agentic Workflows and Enterprise Business Value. I specialize in designing autonomous systems that don't just "chat," but execute complex business logic.
I focus on moving beyond basic RAG into Agentic Orchestration, focusing on:
- Multi-Agent Systems: Task decomposition and collaborative workflows.
- Human-in-the-loop (HITL): Designing safe, governed AI interactions.
- Self-Correction: Agents that verify their own code and data outputs.
- Tool-Use (Function Calling): Connecting LLMs to ERPs, CRMs, and legacy APIs.
| Category | Tools & Frameworks |
|---|---|
| Orchestration | LangGraph, CrewAI, PydanticAI, Semantic Kernel |
| LLMs & Infra | OpenAI, Anthropic (Claude 3.5), Azure AI Studio, AWS Bedrock |
| Vector & Data | Pinecone, Weaviate, Neo4j (GraphRAG), MongoDB |
| DevOps/Ops | LangSmith (Tracing), Docker, Helicone, Weights & Biases |
The Problem: Sales teams spend 40+ hours manually responding to technical RFPs.
The Solution: A collaborative agent swarm (Legal Agent, Security Agent, and Writer Agent) that pulls from a verified knowledge base to draft compliant responses.
Key Tech: CrewAI, ChromaDB, Streamlit.
The Problem: Tier-1 support tickets fail when APIs return unexpected schema changes.
The Solution: An agentic workflow that detects API errors, references documentation, and "self-corrects" the request payload without human intervention.
Key Tech: LangGraph, GPT-4o Function Calling, Python.
- Cost Optimization: Implementation of token-usage tracking and prompt caching.
- Evaluation: Using RAGAS and custom test suites to ensure 95%+ accuracy in production.
- Security: Designing "Prompt Injection" guardrails for enterprise deployments.
- LinkedIn: linkedin.com/in/samuelegabbio
“The goal isn't to build a chatbot; it's to build a digital employee.”
