Onyi

David's AI Assistant

Hi, I'm Onyi!

I'm David's AI assistant. Ask me about his projects, skills, experience, or anything else you'd like to know!

Experience

My professional journey and career highlights

2025-09-01 - Present

AI/ML Research Assistant

MINDS Lab, Minnesota State University • Mankato, MN

  • Researching multimodal AI and scalable ML systems under Dr. Naseef Mansoor.
  • Co-authored 1 published paper (CADSCOM 2024), 1 accepted (CAI 2026), and developing a paper targeting AAAI on semi-structured data generalization.
  • An Agentic System for Schema-Aware NL2SQL Generation – Accepted, CAI 2026. Multi-agent system achieving 50.87% execution accuracy on BIRD benchmark.
  • Evaluating Credit Risk Using Interpretable ML Models – Published CADSCOM 2024; Accepted AMLDS 2025.
2026-01-19 - 2026-04-17

Data Scientist, AI Infrastructure

ProjectPath.AI • Boston, MA (Remote)

  • Shipped an AI project assistant with corrective RAG (CRAG) and LLM function calling with 9+ tools, enabling users to query project state, receive real-time guidance, and perform CRUD operations directly through chat.
  • Engineered 10+ health signals (blocked-task rate, velocity delta, stakeholder activity) powering a transition readiness framework that replaced manual project assessments with evidence-based scoring.
  • Architected a Stage Intelligence Engine using semantic task classification and signal-based heuristics across 28 stages and 4 path variants, lifting classification confidence from 0.70 to 0.90.
2025-05-19 - 2025-08-08

Business Intelligence Engineer Intern

Amazon, RME - Decision Science Team (DST) • Bellevue, WA

  • Built an end-to-end ETL pipeline (Python, SQL, AWS Glue, S3, Redshift, CDK) ingesting operational data from 97+ North American fulfillment sites and classifying technician actions via Bedrock LLM integration, replacing manual triage that could not scale to pipeline volume.
  • Shipped a QuickSight dashboard tracking Resolution Rate, Actionability Rate, and True Positive Rate for robotic miss-sort alerts across 97+ ARS sites, giving operations leaders site-level visibility into alert effectiveness for the first time and moving intervention decisions from gut feel to data.
  • Developed statistical anomaly detection on solar energy data, surfacing hidden cost inefficiencies and translating findings into executive-ready recommendations adopted by the team for next-step modeling.
  • Pioneered Amazon Managed Grafana rollout for DS Team: configured SAML auth and migrated Advanced Refrigeration Monitoring (ARM) dashboards for Amazon Fresh sites from QuickSight SPICE scheduled refresh to Grafana near real-time observability.
2024-09-23 - 2024-12-20

Database Administrator Intern

Northflow Solutions, Inc. • North Mankato, MN

  • Optimized Oracle PL/SQL queries via index tuning and schema redesign, improving report generation performance by 30% and reducing analyst wait times.
  • Built a natural language-to-SQL interface enabling non-technical users to self-serve queries, reducing analyst request turnaround from hours to minutes.
  • Built a RAG-based knowledge retrieval tool integrating LLMs with enterprise documentation on AWS Bedrock, cutting developer search time by an estimated 70%.
2022-01-03 - 2023-07-21

Data Analyst

Kenya Revenue Authority • Nakuru, Kenya

  • Automated ETL processes for daily revenue reporting (Python, Excel Power Query), eliminating 3 hrs/day of manual effort and improving forecast accuracy for revenue compliance officers across the South Rift Valley region.
  • Designed dashboards integrating 5+ data sources that became the executive reporting standard; raised the data utilization index from 73% to 75% through data governance initiatives.