Agentic Trading
A breakout agent trading a live or simulated feed — signal engine, deterministic guardrails, paper fills, and a kill switch.
Hands-on guides plus four interactive demos you can run right now — semantic search, a RAG chatbot, a live MCP client/server, and an agentic trading loop — built in the pvakati-tech.ai lab.
// ai_catalog_search
Synthetic 10k-product catalog. Experimentation only — not medical advice.
A breakout agent trading a live or simulated feed — signal engine, deterministic guardrails, paper fills, and a kill switch.
A real MCP client & server speaking JSON-RPC 2.0 in your browser — watch the handshake, then invoke real tools.
Semantic search over a synthetic 10k-product catalog with Pinecone — filters, relevance scores, and explainability.
A grounded RAG chatbot over a support FAQ with citations, moderation on every input, and swappable data.
Production retrieval-augmented generation end to end — ingest, chunk, embed, retrieve, rerank, generate, and evaluate.
MACH principles and a strangler-pattern path from a SAP Hybris monolith to composable, headless commerce.
Offline eval sets, LLM-as-judge, human review, online metrics, and regression suites that keep agents shippable.
A layered blueprint and reference flows for planners, tools, memory, guardrails, and multi-agent orchestration.
Controlled-automation scenarios across shopping, merchandising, service, and fulfillment — plus healthcare drug commerce.
Securing AI agents with OAuth for people, scoped JWTs for agents, RBAC + policy for every write, and audit logs.
A decision framework for choosing the right auth pattern across AI apps, agents, workers, and provider integrations.
Scenario-tuned recommendations, fit scores, and docker-compose starters for the top vector databases.