Services

Applied AI systems your team can run in production

I help engineering and research teams ship document-grounded assistants, evaluation pipelines, and monitoring platforms—backed by SemEval research and real deployments.

  • SemEval 2023–2025
  • IEEE C3 2025
  • VerbaNexAI Lab · UTB
  • Live SGM dashboards
Start a conversation See featured project

What I offer

Conversational AI & RAG systems

For: Teams drowning in technical or operational documentation

Document-grounded assistants with private or on-prem LLM options and production APIs—so answers stay tied to your corpus.

What's included

  • RAG over manuals, SOPs, and internal docs
  • Private / on-premise LLM setups when data cannot leave
  • FastAPI backends and workflow automation (e.g. n8n)

Dataset research, evaluation & creation

For: Research and product teams that need reliable data, not only models

Corpus design, curation, and evaluation pipelines so datasets are fit for shared tasks, training, or quality gates.

What's included

  • Corpus design for research or shared tasks
  • Data quality and suitability analysis
  • Evaluation pipelines (hallucination / labeling consistency)

AI architecture for monitoring platforms

For: Organizations running multi-site sensors and early-warning flows

End-to-end AI/data architecture for ingestion, analytics hooks, and alerting—designed to scale across sites.

What's included

  • Environmental or hydraulic monitoring architectures
  • Real-time ingestion and alert design
  • Requirements and design for multi-city deployments

IoT & meteorological data platforms

For: Operators who need station-level boards they can trust daily

Collect, store, and visualize high-frequency meteorological and IoT data with APIs and operator dashboards.

What's included

  • TimescaleDB / time-series storage and ETL
  • REST APIs and operational dashboards
  • Station networks and board-style operator views

How we work

  1. 1

    Discover

    Clarify goals, data sources, constraints (privacy, latency, on-prem), and success criteria.

  2. 2

    Design

    Agree on architecture, evaluation approach, and a thin vertical that can ship first.

  3. 3

    Build

    Implement APIs, pipelines, and UIs with iterative demos—not a big-bang handoff.

  4. 4

    Handoff

    Document, deploy, and transfer ownership so your team can operate and extend the system.

Common questions

Do you work remotely?

Yes. Most engagements are remote with clear async updates; on-site is possible in Colombia when needed.

Can you work with private or on-premise LLMs?

Yes. RAG and assistant work can stay on your network when documents or models cannot leave the organization.

How do projects usually start?

A short conversation to scope the problem, then a focused first milestone (often a working vertical in weeks, not months).

Is this only research, or production too?

Both. Shared-task research informs the methods; production work (APIs, dashboards, monitoring) is what I ship for teams.

Ready to scope a RAG or monitoring project?

Tell me about your use case—NLP assistants, datasets, or IoT monitoring. I usually reply within 1–2 business days.

Email Anderson