Data & AI
Data and time-series pipelines, search and indexing, document RAG and agent architecture design.
What it involves
On the data side: time-series pipelines, InfluxDB, OpenSearch, indexing, aggregation, search and analytics.
On the RAG side: document ingestion, Apache Tika / OCR, chunking, embeddings, Qdrant, then a local language model run through Ollama for answers with sources.
Capabilities
LLMs that do more than
answer.
I work on agent architectures that allow a language model to act on tools and external systems, chain multiple steps and retain the context needed to complete a task.
The goal is to build controllable, observable agents: tool access through explicit interfaces, state management, validation of sensitive actions, and logging of actions, tool calls, results and errors, with recovery after failure.
Explicit actions
Limited permissions
Validation of sensitive operations
State
Memory
Retrieval
Tool calling
MCP
Multi-step workflows
Logs
Tracing
Errors
Retries
Human intervention