Capability
Data Science
Decisions are only as good as the analysis behind them. We bring statistical rigor, experimentation, and clear communication to your data — producing insights you can act on and defend.
Proven where it counts
Real-time insights across unified systems
Built secure data pipelines connecting Salesforce cloud and on-prem TSA systems, giving decision-makers real-time, actionable insights.
- Salesforce
- Security
- Data Pipelines
High-performance data engineering
Built and supported a data warehouse capable of sub-second, high-throughput queries. Designed ETL pipelines using Apache Spark Structured Streaming in the cloud.
- Apache Spark
- ETL
- Big Data
API design & development
Accelerated delivery with cost savings for two public-facing API products. Transitioned from Akamai to AWS firewall for a direct cost reduction.
- API
- AWS
- Cost Optimization
What we deliver
Statistical Modeling
Rigorous models grounded in statistical integrity — built to be explained, challenged, and defended, not just to fit.
Experimentation & Causal Inference
A/B tests, quasi-experiments, and causal methods that separate what worked from what merely happened.
Forecasting
Demand, workload, and risk forecasting with honest uncertainty intervals — so plans survive contact with reality.
Fraud & Anomaly Analytics
Detection models for fraud, waste, and abuse — improving compliance and recovering value from vast datasets.
Decision Analytics & Visualization
Dashboards and decision tools that put the right number in front of the right person at the right time.
Reproducible Research
Versioned data, code, and analysis so every finding can be re-run, audited, and trusted long after delivery.
Approach
Statistical integrity over statistical theater.
It is easy to produce a chart; it is harder to produce a conclusion that holds. We are explicit about assumptions, honest about uncertainty, and disciplined about reproducibility — because in mission-critical settings, an overconfident answer is worse than no answer.
- Methods matched to the question, not the fashion
- Uncertainty quantified and communicated plainly
- Every analysis versioned and reproducible
- Findings delivered as decisions, not decks
120-day MVP delivery. Partnered with a prime contractor to rapidly deliver an MVP web application for audit reporting — from question to shipped product in 120 days.
Technologies we work in
- Python
- R
- SQL
- pandas
- scikit-learn
- statsmodels
- Jupyter
- Streamlit
- Tableau
- Power BI
Common questions
How is data science different from AI/ML here?
Our AI/ML practice builds production learning systems; our data science practice answers questions — with statistics, experiments, and models whose job is insight and defensible decisions. They share foundations, and on many engagements they work together: analysis identifies the opportunity, ML operationalizes it.
What does "defensible" analysis mean for a federal agency?
It means every number can survive scrutiny: documented assumptions, versioned data and code, quantified uncertainty, and methods appropriate to the question. When a finding informs policy, budget, or an IG review, you can show exactly how it was produced — and reproduce it on demand.
Can you work with sensitive or regulated data?
Yes. We operate inside your security boundary and compliance framework — including NIST 800-53 and FISMA environments — with privacy-preserving practices such as minimization, de-identification, and role-based access built into the analysis workflow.