AV

Deployments & Product Work

Tools, models, and systems I've helped ship.

From internal hubs in consulting to ML capstones and data products, these are the things that moved beyond slides into working artifacts.

Deal & Internal Tools

Decision-support tools built inside consulting and deal environments, where speed and explainability matter as much as accuracy.

INTERNAL TOOL · DEPLOYED

Deal Scenario Analysis Hub

Internal web-based tool to quickly simulate standalone, synergy, and carve-out scenarios across deals, giving partners and teams a single place to reason about outcomes.

  • • Aggregated and normalized financial and operational inputs.
  • • Framed outputs around questions partners actually ask in deal rooms.
  • • Iterated based on feedback from teams using it across engagements.

INTERNAL TOOL · METRICS HUB

Business Reporting & Sector Metrics Tool

Metrics hub combining internal project data and market benchmarks to make it easier for teams to compare companies, sectors, and deal theses without rebuilding the same views from scratch.

  • • Defined a common schema for sector- and deal-level metrics.
  • • Built views that could be reused across pitches and diligence work.
  • • Focused on reducing time from question to “good enough” answer.

ANALYTICS · BENCHMARKING

Benchmarking & Performance Comparison Tool

Framework and templates to benchmark companies against peers across margin, capital intensity, and operational metrics, supporting both diligence and portfolio work.

  • • Standardized metric definitions to reduce debate about the numbers.
  • • Built reusable views aligned to how investors talk about businesses.
  • • Enabled faster, more consistent comparisons across targets.

OPERATIONS · NETWORK MODEL

A&M Trucking Network & Cost Model

Operational and financial model for a trucking / logistics network, tying route patterns, utilization, and cost structure to concrete levers for performance improvement.

  • • Mapped routes, lanes, and asset utilization into a single structure.
  • • Linked operational changes directly to unit economics.
  • • Helped teams see which levers had real impact on cost per mile.

Data Products & In-Progress Builds

Experiments where I'm closer to the metal — designing schemas, writing SQL/Python, and thinking about how operators would actually use the output.

DATA PRODUCT · IN PROGRESS

MyTab — Restaurant Tip & Performance Analytics

Prototype data product to quantify server performance and design a fair, incentive-aligned tip pool — giving operators a clearer picture of who drives revenue and why.

  • • Synthetic PoS schema in PostgreSQL for tickets, tips, and shifts.
  • • Python-based analysis for revenue, tip distribution, and fairness metrics.
  • • Designed around metrics that feel explainable and fair to staff.

ANALYTICS · DEFENSE / INDUSTRIALS

FDS Project — Defense & Industrial Analytics

Analytical work exploring defense and industrial spend, focused on building a structured view of segments, players, and where value is likely to accrue.

  • • Organized messy data into a usable, queryable structure.
  • • Framed outputs around investor- and operator-relevant questions.
  • • Served as a sandbox for further ML / forecasting ideas.

Academic & ML Projects

Work from MIT and Northeastern where the output was a working model or pipeline, not just a paper — with an eye toward how these ideas could live in the real world.

CAPSTONE · COMPUTER VISION

Facial Expression-Based Pain Assessment (Engineering Capstone)

Spearheaded a remote pain assessment capstone using machine learning to adapt to COVID-19's constraints on in-person care. The team captured facial expression data remotely via Zoom and used OpenFace 2.0 to extract facial action units (AUs), training an SVM model to estimate pain levels.

  • • Defined project scope and methodology in collaboration with advisors.
  • • Used OpenFace 2.0 to convert video into structured AU features.
  • • Trained an SVM model whose performance improves with new data over time.

MIT BOOTCAMP · CNN

Malaria Detection Capstone (MIT Python for AI & Data)

End-to-end computer vision pipeline to detect malaria from blood smear images using convolutional neural networks, with attention to data quality, model performance, and potential deployment constraints in low-resource settings.

  • • Built data loaders, augmentations, and training loops in Python.
  • • Tuned CNN architectures and evaluated against baselines.
  • • Considered how such a system might integrate with real clinical workflows.

ML PROJECT · VISION

Street View House Number Digit Recognition

Recognized street view house number digits using artificial and convolutional neural networks on the SVHN dataset, exploring the tradeoffs between simpler models and deeper CNNs.

  • • Implemented and compared basic ANNs and CNNs.
  • • Experimented with hyperparameters, regularization, and augmentation.
  • • Focused on interpretability and where the models failed.

ANALYTICS · PYTHON / STATS

FoodHub Order Analysis

Analytical project for a food aggregator, using Python to understand order patterns, customer behavior, and platform dynamics through EDA and statistics.

  • • Performed extensive EDA on order and customer-level data.
  • • Used statistical tests to validate hypotheses about demand patterns.
  • • Translated findings into simple, decision-ready recommendations.