dhwanilmori.com
portfolio  /  2026

Dhwanil Mori.

AI ResearcherFounderMS Data Science · GWU
Arlington, VA  ·  available for select collaborations --:--:-- EST
↓ scroll
est. 2025 —  v.01
// profile
01 — ABOUT i. quiet confidence

I build AI systems that fail gracefully and reason collectively. My work spans multi-agent LLM research, intelligent procurement platforms, and applied machine learning — from academic manuscripts to production deployments.

Currently: MS Data Science at GWU, Founder of RAIN (Anthropic-backed), and Research Collaborator with Prof. Neil F. Johnson.

dhwanilmori@gmail.com · GitHub ↗ · LinkedIn ↗ · Scholar ↗
// work
02 — EXPERIENCE 06 roles

Founder

RAIN · Backed by Anthropic

Building agentic retail intelligence — a multi-agent system that negotiates, forecasts, and reasons over procurement at the speed of commerce.

AI multi-agent retail-tech
Dec 2025 — Present

Head of Research Design

HeartWise

Designing experimentation infrastructure and data pipelines for clinical health behavior research.

A/B testing data pipelines
Jan 2026 — Present

Data Analyst

Resilience Inc.

Building reporting and analytics workflows for nonprofit operations and impact measurement.

SQL Python nonprofit
Feb 2026 — Present

Research Assistant

GWU Physics · LLM Council Lab

Investigating reinforcement-learned coordination protocols across heterogeneous LLM agents. IEEE manuscript in preparation.

LLM RL IEEE
Nov 2025 — Present

Research Assistant

GWU Corcoran Arts

Generative-AI pipelines for cross-modal artistic research; HPC orchestration on AWS for diffusion + audio models.

HPC GenAI AWS
May 2025 — Jan 2026

Research Collaborator

GWU Physics · AI Failures Group

Studied adversarial failure modes in open-source language models — emergent collusion, jailbreak surface, and graceful degradation.

adversarial-AI open-source models
Sep — Dec 2025
Download Full CV ↓
// research
03 — PUBLICATIONS 03 papers
P.01

Safe and Efficient Resource Allocation in Multi-AI-Agent LLM Systems.

Under review · JAIAI & IEEE, 2026
w/ Prof. Neil F. Johnson
★ Featured in Button-down AGI newsletter

A framework for cooperative resource arbitration across heterogeneous LLM agents — introducing failure-graceful negotiation protocols with provable safety bounds.

P.02

Temperature-Induced Symbolic Phase Transitions in Autoregressive Language Models.

GW CCAS Research Showcase · April 2026
w/ Prof. Neil F. Johnson + 22 co-authors

Empirical evidence of discrete reasoning regimes in LLMs as a function of sampling temperature — with implications for prompt engineering and alignment.

P.03

Forecasting When Deliveries Fail: A Practical ML Approach for Procurement.

GW CCAS Research Showcase · April 2026
w/ Prof. Junjun Yin

A gradient-boosted classifier for delivery-failure risk in supply chains, trained on 2.4M procurement records — deployed in pilot with a Fortune 500 partner.

// recognition
04 — AWARDS 02 honors

GW Trustworthy AI Hackathon

Three prizes: Most Innovative Use of AI, Longest AI Prompt, and Most Pull Requests.

★ Featured in GWU Data Science Newsletter
2026

Google Developer Startup Day

Won for Onboard AI, an intelligent onboarding automation platform. Pitched at GWU Build with AI Day.

★ Selected showcase, Build with AI Day
2025
// stack
05 — TOOLING no progress bars
languages
Python R SQL TensorFlow Scikit-learn XGBoost LightGBM
ai / research
Multi-Agent LLM Claude API Reinforcement Learning NLP A/B Testing
infra / data
AWS GCP Vercel Upstash / Redis Spark Hadoop Neo4j MySQL n8n
analytics / misc
Tableau Power BI HPC / Slurm ComfyUI MusicGen
viewers 500