Every project, with the case study behind it.

Zasti Agentic Automation Platform

A shared agent harness (LangGraph through MCP) with four automation pipelines built on it — sales, resource tracking, marketing, and internal software engineering — adopted company-wide including leadership, with a human approval gate before any consequential action.

Zasti Inc · Design and implementation

4

verticals automated

1

shared agent harness

10

person company, adopted end to end

0

consequential actions without human approval

Unwind

A reversibility layer for agentic tool use. A transparent MCP proxy that works out which agent actions can be taken back, quietly reverses the ones that go wrong, and interrupts you only for the ones that genuinely can't be undone.

Sole author · v0.1.0 · early beta · public since Aug 2026

R0–R4

reversibility classes

3

distribution channels

v0.1.0

early beta release

Apache-2.0

license

verifydoc — per-field selective risk control

Reference implementation for “Valid Per-Field Selective Risk Control for Document Extraction: Three Failure Modes, a Validity Ladder, and When Conditioning Pays.” It shows that the natural per-field accept/review procedure silently violates its own selective-risk guarantee on real documents.

Sole author · Submitted 28 July 2026 · announced August 2026 · arXiv preprint · 14 pages

13,859

fields evaluated

800

CORD receipts

3

failure modes diagnosed

49.0%

base field accuracy

agenthallu-bench — auditing agent evaluation

Code and data for “Auditing Automated Evaluation, Error Propagation, and Runtime Mitigation in Tool-Using Language Agents.” It audits the widely-assumed reliability of automated evaluation for tool-using LLM agents against human annotation.

Sole author · v1 April 2026 · revised 9 August 2026 · arXiv preprint · 11 pages, 4 figures, 8 tables

14,750

execution traces

13

LLM agents audited

4

task domains

9 + 4

proprietary + open-weight models

Autonomous Multi-Agent Research Platform

A Planner–Retriever–Reasoner–Validator agent loop with persistent memory and tool use across PubMed, ArXiv, and Semantic Scholar APIs.

+41%

synthesis quality

−68%

hallucinated citations

5K+

concurrent sessions

<200ms

median latency

Multimodal Clinical AI Assistant

GPT-4V vision, FHIR parsing, Neo4j patient graphs, and vector search combined for knowledge-graph patient profiling.

94%

retrieval accuracy

15+

biomarker categories

87%

anomaly sensitivity

3

pilot healthcare teams

Conversational AI Search Engine

QLoRA-fine-tuned Llama 3.1 with hybrid dense + sparse retrieval and automated citation and fact-verification.

−40%

misinformation

150ms

P95 time-to-first-token

1K+

concurrent users

+35%

NPS

EEG–fMRI Fusion for Neural Decoding

A multimodal neuroimaging framework predicting visual stimuli by fusing 70-channel EEG with 3T fMRI (Wakeman–Henson dataset) using temporal convolutional networks.

University of Cincinnati · M.S. thesis · advisor Prof. Vikram Ravindra · 2022 — 2024

84.8%

within-subject accuracy

81.1%

cross-subject (LOSO)

0.93

ROC-AUC (within)

0.90

ROC-AUC (cross)