I build decision-critical AI systems with strong reliability, observability, and operational rigor.
Operating model
I build production decision systems for operational workflows. I use LLMs when they improve a specific step, but I default to deterministic methods when consistency, latency, and control matter more.
How I Build Decision Systems
- Start from decision contracts, constraints, and failure modes before selecting model architecture.
- Combine optimization, forecasting, and agentic orchestration only when each component has a clear operational role.
- Design for operator trust with explicit guardrails, traceability, and human approval on high-impact actions.
What Has To Hold In Production
- Stable behavior under distribution shift and changing operating context.
- Fast diagnosis and controlled recovery when systems fail.
- Outputs that operators can audit and act on with confidence.
Core Methods
Decision SystemsOptimizationForecastingAgent OrchestrationBacktestingModel RankingGuardrailsObservabilityFastAPI Microservices
Employment
Ascentt
Principal AI Engineer I
- Built a reactive allocation arbitrator for post-scramble conflicts using multi-agent orchestration and deterministic optimization.
- Added human-in-the-loop approval gates and real-time decision traces to keep execution controllable and auditable.
- Architected a cloud-agnostic demand forecasting platform across 33 vehicle series using FastAPI microservices and open MLOps tooling.
- Introduced stability-based model ranking and rolling-window probabilistic backtesting to improve planning reliability under changing conditions.
Royal Cyber
ML Engineer
- Built a multi-agent RAG workflow for e-commerce decision support.
- Shipped low-latency inference services and CI/CD for production delivery.
Microsoft
AI Intern - LLM Optimization
- Improved GPT-4 summarization quality through prompt iteration and evaluation loops.
PowerKiosk
ML Engineer
- Applied causal inference and gradient-boosting models for pricing decisions.
Data Science Institute, UChicago
Data Scientist
- Fine-tuned NLP and CV models for policy and aquaculture analytics.
Amazon
Supply Chain Analytics Engineer
- Built forecasting pipelines and analytics workflows that reduced stockouts by about 10%.
Education
The University of Chicago
M.S. in Applied Data Science
Sathyabama Institute of Science and Technology
B.E. in Electronics & Communication