-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
204 lines (191 loc) · 10.2 KB
/
index.html
File metadata and controls
204 lines (191 loc) · 10.2 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Todd Espy — Senior Software Architect & AI/ML Engineer</title>
<link rel="stylesheet" href="styles.css" />
</head>
<body>
<div class="layout">
<main class="content">
<div class="banner">
<h1>Todd Espy</h1>
<p>Senior Software Architect & AI/ML Engineer</p>
</div>
<p>
Senior architect with deep production roots in enterprise systems — contact center
platforms, distributed middleware, large-scale integrations. Started in AI before
the hardware existed to support it: fuzzy logic and genetic algorithms, published
at IEEE and NASA conferences. Picked up a couple of patents in speech recognition
along the way. Currently focused on algorithmic trading infrastructure and
multi-agent AI pipelines, both running on local hardware.
</p>
<p>
Twelve years at Verizon as technical lead on large-scale, mission-critical systems:
IVR consolidation for Kaiser Permanente (100K calls/hour), Dialogflow CX chatbot
deployments for state government agencies, mobile SDK development, and enterprise
tooling adopted organization-wide.
</p>
<p>
M.S.E.E., Duke University.
</p>
<h2>Open Source Projects</h2>
<div class="project">
<h2>
<a class="contact-link" href="lean-optimizer.html">lean-optimizer — Parallel Optimization Engine
for Algorithmic Trading</a>
</h2>
<p>
Production-grade parallel optimization engine for QuantConnect LEAN algorithmic
trading strategies. Compiles a C# strategy once, spins up a pool of warm Docker
containers with persistent .NET harness processes, and searches the parameter space
using an incremental genetic algorithm. Additional pipeline stages (LHS,
Bayesian/Optuna TPE, Local Grid) are available and configurable.
</p>
<p>
Measured head-to-head against LEAN CLI's built-in grid search on a 679,140-point
parameter space: the GA converged in 611 evaluations at 1.8x higher throughput per
worker — completing in under 3 minutes what grid search would take ~88 hours.
</p>
<div class="stats">
<div class="stat">
<span class="stat-value">1,112x</span>
<span class="stat-label">fewer evaluations</span>
</div>
<div class="stat">
<span class="stat-value"><3 min</span>
<span class="stat-label">vs ~88 hr grid search</span>
</div>
<div class="stat">
<span class="stat-value">177</span>
<span class="stat-label">passing tests</span>
</div>
</div>
<div class="tags">
<span class="tag">Python</span>
<span class="tag">C# / .NET</span>
<span class="tag">Docker</span>
<span class="tag">Optuna</span>
<span class="tag">Genetic Algorithm</span>
<span class="tag">QuantConnect LEAN</span>
<span class="tag">DGX Spark</span>
</div>
<p class="repo-note">
<a class="contact-link" href="https://github.com/t-espy/lean-optimizer-public"
target="_blank">github.com/t-espy/lean-optimizer-public</a>
</p>
</div>
<div class="project">
<h2>
<a class="contact-link" href="https://github.com/t-espy/langgraph-factory"
target="_blank">langgraph-factory — Multi-Agent Code Generation with Local LLMs</a>
</h2>
<p>
Multi-agent code generation pipeline using LangGraph and locally-hosted LLMs via
Docker Model Runner. A foreman model (gpt-oss 20B) handles architecture planning
and code review; a coder model (Qwen3 Coder 80B) handles file generation and fixes.
Automated build-fix-regenerate loop recovers from failures without human intervention.
Produces buildable Next.js App Router projects from a natural-language spec. Runs
entirely on local hardware with no external API dependencies.
</p>
<p>
Includes a cost and data privacy analysis comparing local inference vs. per-token
cloud APIs — practical reference material for teams evaluating on-premise LLM deployment.
</p>
<div class="tags">
<span class="tag">LangGraph</span>
<span class="tag">Docker Model Runner</span>
<span class="tag">gpt-oss 20B</span>
<span class="tag">Qwen3 Coder 80B</span>
<span class="tag">Next.js</span>
<span class="tag">Local LLM</span>
<span class="tag">Multi-Agent</span>
</div>
<p class="repo-note">
<a class="contact-link" href="https://github.com/t-espy/langgraph-factory"
target="_blank">github.com/t-espy/langgraph-factory</a>
</p>
</div>
<div class="project">
<h2>
<a class="contact-link" href="https://www.route-forge.com" target="_blank">Route Forge — SaaS
Multi-Stop Route Planning</a>
</h2>
<p>
End-to-end SaaS web application for multi-stop route planning. Built on Next.js,
React, Tailwind, Supabase PostgreSQL, and Stripe, hosted on Vercel with automated
GitHub Actions CI/CD. Architected as a three-app system using the BFF pattern with
cross-subdomain authentication. Demonstrates full production deployment from zero:
payments, auth, SEO, and go-to-market infrastructure.
</p>
<div class="tags">
<span class="tag">Next.js</span>
<span class="tag">React</span>
<span class="tag">Supabase</span>
<span class="tag">Stripe</span>
<span class="tag">Vercel</span>
<span class="tag">GitHub Actions</span>
</div>
</div>
<h2>Publications & Patents</h2>
<ul>
<li>
<a class="contact-link" href="https://ieeexplore.ieee.org/document/258768" target="_blank">Fuzzy
Logic Control of AC Induction Motors</a>
— IEEE International Conference on Fuzzy Systems (co-author)
</li>
<li>
<a class="contact-link" href="https://ntrs.nasa.gov/citations/19930013020"
target="_blank">Application of Genetic Algorithms to Tuning Fuzzy Control Systems</a>
— NASA SOAR Conference & Third International Workshop on Neural Networks and Fuzzy Logic
(co-author)
</li>
<li>
<a class="contact-link" href="https://patents.google.com/patent/US8639505" target="_blank">US
8,639,505</a>
— Method and systems for simplifying copying and pasting transcriptions from a dictation-based
speech-to-text system (co-inventor)
</li>
<li>
<a class="contact-link" href="https://patents.google.com/patent/US8639512" target="_blank">US
8,639,512</a>
— Method and systems for measuring user performance with speech-to-text conversion for dictation
systems (co-inventor)
</li>
</ul>
</main>
<aside class="sidebar" id="contact">
<h2>Contact</h2>
<p>Feel free to reach out:</p>
<ul style="list-style: none; padding-left: 0;">
<li><strong>Email:</strong> <span id="email"></span></li>
<li><strong>LinkedIn:</strong> <a href="https://www.linkedin.com/in/toddespy/"
target="_blank">linkedin.com/in/toddespy</a></li>
</ul>
<h2 style="margin-top: 2rem;">Certifications</h2>
<div class="cert-badges">
<a href="https://www.credly.com/badges/0aa4f899-e591-4644-96ed-97cb862f94ab/public_url" target="_blank">
<img src="images/aws-certified-machine-learning-engineer-associate-e.png"
alt="AWS Certified ML Engineer" width="180" />
</a>
<a href="https://www.credly.com/badges/ef070385-9cd4-4596-8289-5796bf4c61e4/public_url" target="_blank">
<img src="images/aws-certified-ai-practitioner-early-adopter600x600.png"
alt="AWS Certified AI Practitioner" width="180" />
</a>
<a href="https://www.credly.com/badges/6e18930b-2286-423c-a406-49bb93acd98d/public_url" target="_blank">
<img src="images/aws-certified-cloud-practitioner.png" alt="AWS Certified Cloud Practitioner"
width="160" />
</a>
</div>
</aside>
</div>
<script>
(function () {
var u = "todd.espy", d = "gmail.com", e = u + "@" + d;
var s = document.getElementById("email");
if (s) s.innerHTML = '<a href="mailto:' + e + '">' + e + '</a>';
})();
</script>
</body>
</html>