Coding Shinobi — portfolio and presentation of 2014–2026

Bar Raiser & Team Praiser

Over twelve years across multiple domains including: medical, geospatial, sensors and simulations, VR/AR/XR, and agent loops with AI/ML Pipelines.

Partnering with and building high performing teams that love what they do.

Scroll to selected work

Selected work

Five I'd show you first.

Surgical imaging that loads in a blink without ever guessing a measurement. Half a million clinical trials mapped for human discovery. High-throughput analyst pipelines paired with shared Unreal worlds. Playable interactive prototypes, and forty federal services unified under one roof. Different domains, same heartbeat: elevate the team, protect the user, and build systems you can trust.

Bone Viewer 3D workspace with a segmented volume, structure list, and pipeline controls Augmenting Orgs

Clinical imaging

Orthopedic 3D imaging

Streaming 3D surgical scans in real time while providing AI/ML analytics on structures.

Live demo available
Open
Map-based clinical trials search interface AI Done Right

Clinical data

Clinical trials discovery

Connecting patients and researchers to 607,160 real studies using AI that is structurally incapable of hallucinating.

607,160 studies mapped● zero hallucination
Open
Networked simulation environment Teams and Platforms

Throughput & simulation

High-velocity physics & digital twins

Turning hours of manual data wrangling into a 1 TB/min automated pipeline—and a 10,000-object Unreal simulation.

64-process HPC · 1 TB/min● zero setup errors
Open
Collectible card game prototype built in Unity Teams and Platforms

Interactive

Interactive Unity/Unreal Experiences

Designing tactile game systems in Unity and coaching engineering teams to bring scientific simulations into XR.

CCG Architecture + XR Sims● shipped
Open
AI Done Right

Platform Architecture

Forty services, one medallion

Unifying 40 fragmented federal services and 100+ relational tables into one governed Medallion platform.

40+ services · 100+ tables● 4× delivery velocity
Open

2014—2020

Early Days

Where the foundation was forged: self-hosted live video before it was a commodity, multi-tenant logistics, edge hardware, and mentoring my first engineers.

Honorable mentions

Also from the early years.

  1. 2016—2017

    Edge-computed dashcam telemetry

    Fused radar, camera vision, and inertial sensors directly on a Raspberry Pi for real-time driver safety analytics—engineered in multi-threaded C++ to run completely offline at the edge.

    C++ · OpenCV · TensorFlow · IPC · Raspberry Pi
  2. 2018

    GeoPackage RTE reference implementation

    Built the living reference implementation for an emerging geospatial routing standard in 3D terrain, presenting it directly to the standards committee.

    React · Cesium · Leaflet
  3. 2019—2022

    3D CT bone deformation pipelines

    Created custom 3D medical volume labeling tools and trained deep-learning segmentation pipelines to help clinical researchers track complex bone deformations.

    Python · VTK · TensorFlow · PyTorch · MONAI · ONNX
  4. 2020—2022

    Air traffic control training simulation

    Real-time, multi-user airspace simulations that synchronized live controller sessions and adapted training scenarios to each student's weak spots.

    React · Redis · Python

2018—2024

Teams and Platforms


From building my own team of ten;
  a fast-paced collaboration with eighty members of an org,
   upgrading the workflows of twelve scientific analysts,
    and becoming a bar raiser that provided technical-creative advice.

Early systems built the learn-it-all foundation.

But solo technical mastery was never the endgame—multiplying the people around me was.

Cross-Org Alignment

AR city twins at scale

Embedded within a ten-person strike team inside an eighty-person defense and geospatial program, bridging the gap between independently built AR hardware and world-scale simulation engines. The real breakthrough wasn't just aligning coordinate math—it was aligning competing engineering cultures into one shared standard.

10-person team · 80-person program

● aligned & shipped

React · Cesium · Unity · Java / Spring

Virtual reality digital twin simulation
Networked digital twinArchive

Human-Centered Automation

Empowering the analysts' floor

Twelve scientific analysts were burning hours hand-wrangling massive 900 GB simulation files, where a single typo could ruin a run. I pulled up a chair right beside them, listened to their pain points, and engineered a 64-process automation suite that ingested 1 TB/min and wiped out setup errors completely—freeing the team to focus on actual science, and paving the way for a live Unreal digital twin.

12 analysts freed · 1 TB/min throughput

● zero setup errors

Python · Spark · Qt · HDF5 · Unreal Engine · C++

Collectible card game prototype built in Unity
Unity CCG prototypeScreenshot

Technical-Creative Advisory

Interactive Unity/Unreal Experiences

Acted as a technical-creative bar raiser for client engineering teams—coaching them on how to translate rigid scientific simulations into intuitive, tactile XR experiences, while sharpening complex state-machine and interaction patterns through playable Unity prototypes.

Client teams upskilled · Playable XR

● built & mentored

Unity · C# · Unreal Engine · C++

Live Operations

Sixteen cameras, one living memory

Venue operations teams were drowning in sixteen simultaneous video feeds with no way to scrub back and connect the dots. I architected a distributed computer-vision pipeline that indexed every detection across space and time—turning chaotic live footage into an instant, searchable timeline.

16 live feeds · real-time spatial index

● production scale

Python · Kubernetes · Kafka · Redis · OpenCV · ffmpeg

2022—2026

Augmenting Orgs

Stepping into high-stakes clinical, enterprise, and legal environments—where speed changes lives, but unverified data is never an option.

Bone Viewer 3D workspace with a segmented volume, structure list, and pipeline controls
Bone Viewer · 3D workspaceScreenshot

Clinical imaging

Precision orthopedic 3D imaging

Surgeons and Orthopedists have always been looking into safer and secure ways of integrating Machine Learning into practices. I re-architected the clinical viewer to allow for compliant transmission and auditable AI/ML pipelines for various tasks.

Real-time slice streaming · Verified voxel scale

● 168 invariant tests

FastAPI · vtk.js · S3 · Terraform · 168 tests

Interface for searching clinical trials
Clinical trials discoveryArchive

Healthcare Discovery

Mapping 607,160 clinical trials

Finding a life-saving clinical trial shouldn't require a degree in database query syntax. I mapped over 607,000 global studies and 3.5 million medical facilities into an interactive geospatial platform—letting patients and clinicians ask complex eligibility questions in plain English while guaranteeing every single result is backed by a verified clinical record.

607,160 studies · 3.5M facilities mapped

● verified records

Next.js · FastAPI · PostgreSQL / PostGIS · Clerk

Enterprise Governance

Petabyte-scale legal discovery

At a Fortune 500 financial institution, legal-hold discovery across a petabyte of sensitive records was bottlenecked by rigid legacy infrastructure. Working closely with security and release teams, I redesigned the sharded storage layout and Microsoft Entra RBAC governance—accelerating case reconstruction and cutting our deployment cycle from three months down to four weeks.

1 PB governed · 3 months → 1 month delivery

● enterprise production

C# / .NET · Blazor · OpenShift · Kubernetes · Entra ID · Purview

2025—2026

AI Done Right

Pioneering verification-first AI architectures—where deterministic truth auditing, multi-tenant security, and zero-trust guardrails are built into the foundation, not bolted on after a failure.

Architectural First

The Deterministic Truthiness Auditor

While much of the industry was shipping raw LLM wrappers and hoping for the best, I architected one of the first multi-tenant financial AI engines with built-in deterministic verification. Every AI proposal is intercepted and audited against strict financial rule engines, tenant security boundaries, and Medallion data lineage before an analyst ever sees a single number.

100% deterministic verification · Multi-tenant RBAC

● zero unverified outputs

Azure Functions · DuckDB · Parquet · Azure SQL · Deterministic Rule Engine

Zero-Trust Grounding

Architecting Out Hallucinations

In healthcare and finance, a "mostly accurate" chatbot is a liability. I designed a zero-trust agent architecture where the model is stripped of free-text generation privileges for factual claims. Instead, it operates strictly through parameterized, schema-locked SQL and geospatial tools—making it structurally impossible to invent a clinical trial, fabricate a metric, or leak cross-tenant data.

Schema-locked tool execution · Zero free-text fabrication

● security by design

FastAPI · PostgreSQL / PostGIS · JSON Schema Guards · Agentic Orchestration

The Blueprint

How we build AI that enterprises actually trust.

  1. Let AI propose; let deterministic code dispose. Never let a probabilistic model be the final authority. Every output passes through a hard, mathematical rule gate before a human ever sees it.
  2. Security is the architecture, not a prompt. Prompt engineering is not a security boundary. Enforce tenant isolation, RBAC, and parameterized tool execution at the infrastructure layer so hallucinations and data leaks are structurally impossible.
  3. Design the escape hatch first. A flashy demo isn't a production system. We track drift, measure validation rejections in real time, and fail safely back to human workflows when confidence drops.
  4. Cryptographic-grade audit trails. Every automated action logs what data was touched, which rule validated it, and which human approved the threshold—turning black-box AI into an auditable asset.

Executive Enablement

Practical AI Agents & Built-In Guardrails

Great architecture only scales when leaders understand how to wield it. I was commissioned to design and teach a 15-episode executive curriculum that demystifies agentic workflows—showing non-technical leaders how to spot high-leverage automation, enforce human-in-the-loop security, and build audit-ready AI governance from day one.

15-episode executive series · Governance & Architecture

● commissioned & shipped

AI Governance · Agentic Workflows · Security Architecture · Executive Advisory

Request the curriculum outline →

Great systems start with the people who build them.

Whether we are aligning an eighty-person engineering org, shaving hours off an analyst's day, or putting hard mathematical guardrails around enterprise AI—I build teams and platforms that hold up under pressure and love what they do.

[email protected] GitHub ↗ V1 — October 2026
1 / 2

Skills

Metrics