Aloysious Kabonge — AI/ML Engineer & Data Scientist

About

From Uganda
to builder.

I came from Uganda to the United States with dreams, questions, and a lot of hope. I graduated from the University of Mary Washington in May 2026 with a B.S. in Data Science — and that degree represents far more than an academic achievement. It represents family sacrifices, community support, and the resilience it took to keep going.

I thrive at the intersection of logic, technology, and real-world impact. My work at SyncData.ai gave me deep experience in LLM evaluation and governance: schema adherence, faithfulness, hallucination control, PII/PHI redaction, prompt-injection resistance, and regression-style testing — building AI that teams can actually trust.

I'm known for being curious, dependable, and collaborative. I learn quickly, ask sharp questions, and communicate technical work in a way that helps teams make better decisions. My guiding mantra: "One day at a time."

PythonFastAPI AWS BedrockClaude API OllamaPrompt Engineering RAG PipelinesLLM Evaluation PostgreSQLPySpark ChromaDBPinecone RabbitMQRedis DockerPower BI DatabricksNext.js Claude Code

Aloysious Kabonge

AI/ML Engineer · Data Scientist

B.S. Data Science — University of Mary Washington, May 2026
Pi Mu Epsilon Mathematics Honor Society
AI Engineering Intern — SyncData.ai (Dec 2025 – Apr 2026)
BI Analyst Intern — Navy Federal Credit Union (Summer 2025)
Data Strategy Intern — banduri / Jade Rabbit AI (Spring 2025)
Resident Assistant, Orientation Leader, Peer Mentor — UMW (4 yrs)
2nd Place — UMW Eagle Egg Pitch Competition (ProofMode)
NCUR 2026 Research Presenter — EAS RAG Pipeline, Richmond VA
SQL Certified — DataCamp (Jul 2025)
Fredericksburg, Virginia 22401 · (540) 419-0472

Experience

Where I've worked.

Built AI-powered compliance automation tooling for regulated data workflows. Engineered a Python/FastAPI ingestion service with MinIO quarantine storage, PostgreSQL metadata persistence, SHA-256 hashing, RabbitMQ event publishing, and idempotent artifact handling. Designed canonical job state machines, Redis worker message contracts, schema validation, and async processing pipelines. Developed LLM enrichment workflows using Ollama and AWS Bedrock with structured JSON outputs, PII/PHI redaction, hallucination checks, prompt versioning, and remediation guardrails. Built executive-ready PDF/HTML compliance reports with AI transparency and traceability. Evaluated LLM models across validity, latency, faithfulness, injection resistance, and factual grounding.

FastAPIAWS Bedrock OllamaPostgreSQL RabbitMQRedis MinIOLLM Evaluation Prompt EngineeringPII Redaction

Analyzed credit and debit card transactional data to understand how Active Duty members' spending behavior changes during Permanent Change of Station (PCS) moves (2019–2024). Built scalable data preparation pipelines in Databricks using PySpark, optimizing joins, aggregations, and feature creation for repeatable reporting. Developed Power BI dashboards surfacing trends and segment comparisons. Delivered stakeholder presentations translating complex analysis into actionable member support recommendations.

PySparkDatabricks Power BIExcel Stakeholder Presentations

Completed banduri's inaugural internship cohort in partnership with UMW College of Business. Contributed to Jade Rabbit, an AI-driven healthcare product. Applied Change Architecture framework to identify critical healthcare pain points and define customer segments. Conducted quantitative and qualitative analyses to validate problem-solution fit. Delivered final presentation to faculty, executives, and stakeholders with actionable product roadmap insights.

Product StrategyHealthcare AI Mixed Methods ResearchStakeholder Comms

Four years of campus leadership across multiple concurrent roles. Served as an RA for 3+ years supporting students in residence halls. Selected as an Orientation Leader for the Class of 2027. Invited to speak at "Let's Talk Internships" sharing lessons from three internship experiences. Active in African Student Union, NAACP, National Society of Collegiate Scholars, and Pi Mu Epsilon Mathematics Honor Society. Volunteered with Habitat for Humanity in Winston-Salem, NC.

LeadershipMentorship Public SpeakingCommunity Building

Projects & Research

Things I've built.

Live Product

🔐

ProofMode

2nd Place · UMW Eagle Egg Pitch Competition

A secure proof-of-process platform for student writing. Students document writing checkpoints, revision reflections, and source notes — generating a tamper-evident PDF with field-level encryption, signed JWT sessions, Argon2 password hashing, and CSRF protection. Built to address academic integrity in an AI-first world without relying on detection tools.

FastAPINext.js PostgreSQLDocker ReportLabJWT Auth Fernet Encryption

Live App ↗ GitHub

Research

📡

EAS RAG Pipeline

Presented at NCUR 2026 · Richmond, VA

An AI pipeline for analyzing Emergency Alerting System regulatory documents. Hybrid retrieval with Pinecone serverless + ChromaDB fallback and automatic external web enrichment via SerpAPI. Grounds LLM responses in trusted source material. Extended into a 911 call analyzer and customer service tool. Originally fine-tuned on GPT-3.5 Turbo.

RAGPinecone ChromaDBOpenAI StreamlitSerpAPI Fine-tuning

GitHub

🤖

AI Alo — Demo Portfolio

3 Production Demos · aialo.io

Three full-stack AI automation tools for local businesses — a restaurant chatbot, real estate lead qualifier with CRM dashboard, and med spa treatment recommender with candidacy screening. Each runs Claude with Ollama fallback, local embeddings, and a provider-facing intelligence dashboard.

FastAPIClaude API ChromaDBOllama RAGLead Scoring

View Demos ↓

Skills

What I bring to the table.

Prompt Engineering RAG Pipelines LLM Evaluation & Governance Claude API AWS Bedrock Ollama (local LLMs) OpenAI API Hallucination Control PII/PHI Redaction Prompt Injection Defense Structured JSON Outputs LLM Fine-tuning Claude Code Agent Frameworks

Python PySpark Databricks Power BI pandas NumPy scikit-learn Statistical Analysis Applied Mathematics Feature Engineering Excel Data Visualization SQL

FastAPI PostgreSQL Docker RabbitMQ Redis MinIO SQLAlchemy REST APIs Async Processing Job State Machines SHA-256 Hashing JWT Auth Next.js

ChromaDB Pinecone sentence-transformers OpenAI Embeddings Hybrid Retrieval Semantic Search Vector Indexing Embedding Pipelines

Communication Collaborative Problem Solving Mentorship & Coaching Public Speaking Stakeholder Presentation Cross-functional Teamwork Intellectual Curiosity Adaptability Critical Thinking Growth Mindset Dependability Community Leadership

Git / GitHub VS Code Claude Code AWS (Bedrock, S3) Render Railway GitHub Pages Streamlit Jupyter Postman SerpAPI ReportLab

AI Alo · Demo Portfolio

Three demos.
Three real pain points solved.

Each demo is a fully functional AI automation app built for a specific local business vertical. Every one runs on Claude (with Ollama as a local fallback), uses local embeddings, and is open source.

🍽️Demo 01

Demo 01 · Restaurant

Casa Alo's Bistro

Fredericksburg, VA · Italian Restaurant

Aria — AI chatbot embedded in restaurant website
RAG over full menu, hours, specials & FAQs
Answers reservation, allergy & dietary questions 24/7
Claude + Ollama local fallback, local embeddings

FastAPIChromaDB ClaudeOllama

GitHub → View Code

🏡Demo 02

Demo 02 · Real Estate

Rappahannock Realty

Fredericksburg, VA · Real Estate Agency

Scout — AI chatbot over 10 listings & 8 neighborhoods
AI lead scoring (1–10) as a non-blocking background task
Lead Intelligence Dashboard — mini-CRM, auto-refresh
In-chat contact capture + personalized follow-up emails

FastAPIChromaDB ClaudeOllama

GitHub → View Code

✨Demo 03

Demo 03 · Med Spa

Luminara Med Spa

Fredericksburg, VA · Medical Aesthetics

Luna — AI treatment recommender by stated concern
Candidacy pre-screener for injectables, laser & peels
Provider Kanban Dashboard with pipeline revenue estimate
Lead scoring by treatment value + draft follow-up emails

FastAPIChromaDB ClaudeOllama

GitHub → View Code

Aloysious
Kabonge.
AI/ML Engineer.

From Uganda
to builder.

Where I've worked.

Things I've built.

What I bring to the table.

Three demos.
Three real pain points solved.

Let's build something.

AloysiousKabonge. AI/ML Engineer.

From Ugandato builder.

Where I've worked.

Things I've built.

What I bring to the table.

Three demos.Three real pain points solved.

Let's build something.

Aloysious
Kabonge.
AI/ML Engineer.

From Uganda
to builder.

Three demos.
Three real pain points solved.