Experience · Enzo Bonelli

Director of AI Engineering

Sears Home Services

Current Apr 2026 — Present San Antonio, TX

Leading the AI engineering organization, owning strategy and architecture for AI-powered customer operations across one of the largest in-home services networks in the country.
Architecting voice AI / conversational AI systems that replace legacy phone-based workflows — real-time speech recognition, LLM-based reasoning and negotiation, and low-latency TTS synthesis over telephony.
Driving a cost-disciplined AI deployment strategy, blending self-hosted open-source models with selective managed services to deliver enterprise capability at predictable, sub-enterprise-SaaS cost.
Setting engineering standards, hiring and mentoring AI engineers, and aligning the AI roadmap with executive priorities.

Voice AI Real-time Streaming LLMs Twilio Whisper TTS Self-hosted Engineering Leadership

Senior Cloud Architect, AI/ML

Caylent (formerly Trek10 — acquired 2025)

Jul 2025 — Mar 2026 San Antonio, TX

Architected and deployed a distributed, air-gapped multi-agent AI tax system (5-agent architecture) with GraphRAG (Neo4j) for traceable reasoning, lineage tracking, and explainable outputs across tax and financial data. Implemented in Python (FastAPI), Dockerized microservices, and a vanilla JS/HTML/CSS frontend — secure, compliant processing of PII inside isolated environments.
Designed and implemented a scalable, serverless AI document-processing platform on AWS (CloudFront, S3, API Gateway, Lambda, Bedrock) for structured and unstructured extraction from PDFs and images. Cut processing costs 20–30% via serverless architecture and model-utilization tuning.
Authored and refined client-facing SOWs and technical proposals, aligning scope, timelines, resourcing, and architecture with executive stakeholder expectations.
Technical lead for cross-functional engineering teams — mentoring developers, running architecture reviews, and enforcing production-grade implementation standards.
Primary technical interface to clients — leading architecture workshops, roadmap discussions, and executive briefings aligning AI strategy with business objectives.

Multi-Agent GraphRAG Neo4j AWS Bedrock Lambda FastAPI Docker Air-Gapped

Artificial Intelligence Solutions Architect

Booz Allen Hamilton

Nov 2023 — Jul 2025 San Antonio, TX

U.S. Space Force (USSF) — AI Lead

AI lead for a $531M USSF program, directing a team developing the data model for the knowledge service and AI integration into USSF infrastructure.
Built agentic RAG/GraphRAG architecture for air-gapped LLM capabilities on IL4+ environments (CrewAI, LangChain, Neo4j, FastAPI, Docker, GitLab).
Partnered with USSF officials and senior leadership to map program requirements for integrating AI into existing DoD infrastructure.
Re-implemented an enterprise-grade API by porting from Python to Go, yielding a more scalable, resource-efficient solution.
Established deliverables, use cases, and timelines for accurate project budgets — personnel and compute cost control.

Internal Research & Development (IRAD) — Software Development Lead / AI SME

Designed an innovative architecture for integrating LLM capabilities into unstructured-data conversion, aligned with ontology standards.
Improved RAG techniques on Neo4j aligned with Commander's Intent, producing optimized data-driven decision support.
Led and mentored junior engineers building a containerized Neo4j graph database, raising team proficiency in graph data management.

CrewAI LangChain Neo4j FastAPI Go Docker GitLab IL4+

Lead Machine Learning Engineer

Booz Allen Hamilton

Apr 2023 — Nov 2023 San Antonio, TX

DoD Chief Digital & AI Office (CDAO) — Technical Lead

Pioneered an LLM-integration architecture that lifted data processing and decision-making capability (Python, LangChain, PostgreSQL, FastAPI, Docker).
Deployed a FAISS vector database and an LLM-integrated FastAPI service, significantly improving analytical precision.
Designed an agent-based approach to assist J2 officers at combatant commands, streamlining critical intelligence operations.
Delivered briefings and demonstrations for Task Force LIMA leadership, keeping the technical work aligned with mission objectives.

U.S. Air Force, F-35 Program — Senior Software Engineer, AI Development & Integration

Designed a data pipeline architecture spanning Postgres, Neo4j, and FAISS.
Built a Docker-containerized microservices architecture for scalability and maintainability.
Used AWS to deploy data pipelines with seamless integration to CUI data sources.
Created a Python middleware API facilitating clean inter-service data exchange.
Engineered an LLM-integration architecture incorporating short- and long-term memory.
Presented progress and recommendations to leadership and stakeholders.

U.S. Army G-38, EOD Knowledge Management — Software Engineer

Trained and deployed an LLM on GCP, demonstrating AI's continuing impact on data engineering and ML for the client.
Deployed and configured a Chroma vector database on GCP — a cost-effective query and organization layer within budget.
Built a PostgreSQL-backed ingestion engine for large-scale ETL across PDFs, audio, and photos.
Mentored engineers in NLP techniques — text classification, keyword extraction, lemmatization, and stemming (NLTK).

LangChain FAISS Chroma PostgreSQL Neo4j FastAPI AWS GCP Docker NLTK

Data Engineer

Booz Allen Hamilton

Feb 2022 — Apr 2023 San Antonio, TX

DoD CDAO — Technical Lead (Deputy)

Served as Deputy Technical Lead, providing technical leadership and overseeing delivery for a team of 3.
Leveraged OCR (pytesseract) in Python for accurate, streamlined ETL.
Mentored junior developers, accelerating their technical growth.

Python pytesseract ETL Mentorship

Data Engineer Intern

iHeartMedia (Internship)

Jun 2021 — Aug 2021 San Antonio, TX

Collaborated on the Analytics Platform team to design and develop a user-friendly interface for cross-functional teams using Docker, Google Cloud Platform, Airflow, SQL, and Python.
Built Python scripts to parse SQL statements and manage Airflow DAGs, incorporating BigQuery operators for seamless data processing.
Worked in a Scrum-based environment, contributing to the software development life cycle and agile team rituals.

Python Airflow BigQuery GCP Docker SQL

Data Engineer

The University of Texas at San Antonio

Jun 2020 — May 2021 San Antonio, TX

Designed and implemented web scrapers to aggregate data on NFL team valuations and individual player statistics — building a foundation for data-driven research.
Cleaned, transformed, and extracted data for rigorous econometric analysis, surfacing actionable patterns.
Built visualizations (graphs, tables, plots) to communicate critical findings to faculty and collaborators.

Python Web Scraping Econometrics Data Visualization ETL

Economic Researcher

New Jersey Department of Human Services (NJDHS)

Jan 2016 — May 2016 Lawrenceville, NJ

Published research

New Jersey Medicaid Expansion: Enrollment, Disenrollment, & Costs

The ACA enabled New Jersey to extend NJ FamilyCare to all non-elderly adults under 138% of the federal poverty level in January 2014. This research examined the Alternative Benefits Program (ABP) — the expansion vehicle that took on roughly 750,000 enrollees in its first 25 months — analyzing who disenrolled, the subset who later experienced a medical emergency, the expenditures associated with those emergencies, and the costs tied to each enrollment path (county offices, general-assistance conversion, the federal ACA portal, and presumptive eligibility).

rider.edu/.../iscap_program_2016_2.pdf

Performed economic analysis of healthcare expansion data, modeling the factors affecting enrollment, disenrollment, and medical-emergency expenditures.
Built econometric models using regression and bivariate analysis on a large NJ state administrative dataset.
Compared cost dynamics across four enrollment paths: county offices, general-assistance conversion, the federal ACA portal, and presumptive eligibility.

Econometrics Regression Analysis Healthcare Policy Medicaid / ACA Published Research