J. Ignacio Esparza

Applied AI engineering · enablement · delivery

José Ignacio Esparza

Turning business needs into working AI tools.

I build agents, data products, cloud workflows, and deployable prototypes, then document and operationalize them so other people can use them with confidence.

View experience Browse projects
  • AI agents & RAG
  • Python & cloud
  • integrations
  • rapid prototyping
  • production handoff

How I work

Practical AI, carried through to adoption.

I work across applied AI, machine learning, automation, and product engineering. My strongest work starts with an unclear operational need: I make the problem concrete, prototype a useful path, test its assumptions, and shape the result into something a colleague can understand and operate.

That includes the less glamorous parts that make AI dependable: connecting services, debugging unfamiliar systems, defining boundaries, documenting decisions, and preparing a clean handoff. I care about technical depth, but the measure of success is whether the system becomes useful to people.

Enterprise experience

Building systems and the conditions around them.

Kyndryl · Applied AI & cloud

From technical prototypes to operational systems.

I contribute to enterprise AI and ML initiatives spanning Python workflows, cloud infrastructure, data and model pipelines, integration, testing, and production handoff. The work requires translating between business context, engineering constraints, and the people who will operate the result.

  • Modular AI/ML workflows and cloud deployment paths
  • System architecture, testing, monitoring, and failure behavior
  • Documentation for engineering, operations, and governance audiences

AI enablement & CoE

Making adoption repeatable.

My current enablement work focuses on the structures that help teams use AI responsibly: practical guidance, tool and use-case evaluation, reusable workflows, governance boundaries, and clear paths from experimentation to supported use.

  • Playbooks and decision support for AI adoption
  • Evaluation, risk, ownership, and lifecycle guidance
  • Materials designed for technical and nontechnical colleagues

Delivery approach

Prototype quickly. Finish deliberately.

I am comfortable entering unfamiliar systems, learning quickly, and moving from a rough request to a working demonstration. I pair speed with traceability so the prototype can be tested, explained, deployed, and improved rather than abandoned.

  • Agents, RAG workflows, APIs, and automation
  • Reproducible builds, tests, and deployment documentation
  • Direct communication and hands-on colleague support

Client names, internal architectures, proprietary datasets, and implementation details are intentionally omitted. Public projects below demonstrate the engineering practices I can discuss openly.

Selected work

Artifacts from the notebook.

Interactive map

CDMX Convenience Map

A static choropleth map for evaluating where to start an apartment search in Mexico City — scoring postal codes and colonias by commute, transit, groceries, gyms, and safety.

  • 1,215 postal codes × 1,837 colonias
  • weighted multi-criteria scoring
  • offline data pipeline (Python + OSM)
  • React + Leaflet, no backend
Try the map View source
Research system

ETF Portfolio Research

A reproducible ETF portfolio research pipeline for optimization, walk-forward backtesting, reporting, and explicit assumptions.

  • explicit assumptions
  • reproducible runs
  • code-driven reports
  • no hand-wavy backtests
Explore the results View source
Android app

Reed RSVP Reader

A speed-reading app for Android built around focus, localization, and privacy-first reading on-device.

  • EPUB / PDF / TXT
  • 100–1000 WPM
  • 4-language localization
  • device-local data
Play Store Privacy policy
LLM workflow

AIIRG

A report-generation workflow for turning open-ended industry questions into structured research outputs with Gemini, search, and document export.

  • research-to-report pipeline
  • prompted analysis structure
  • document generation workflow
View repo

Open questions

Domains under observation.

Machine learning

From classic supervised learning to more experimental modeling paths.

Biology & systems

Complex behavior emerging from local interactions, constraints, and time.

Markets & decision-making

Uncertainty, incentives, and model limits made visible enough to reason with.

Language & learning

How minds absorb, encode, revise, and transform information over time.

Connect

Let us build something useful.

ignacio.ireta@outlook.com linkedin.com/in/ignacio-ireta github.com/ignacio-ireta