session.start — you've reached mark

Agents do my busywork.

because I spend my days and most nights building them.

I'm Mark — an applied AI application developer at Vanguard. I build multi-agent systems inside one of the world's largest asset managers, then go home and build the tools those agents wish they had. Scroll, and you'll meet me the way my agents did: one verified fact at a time.

[source: resume]
[source: github]
[source: mark himself]

Malvern, PARIT Software Engineering '24AWS Certified ×2

see the artifacts start a handshake

trace — how i got here

Read the log, not the résumé.

Every line below actually shipped. Timestamps included.

Aug 2024 – Present

Applied AI Application Developer

Present

Vanguard · Malvern, PA

Designed and implemented a multi-agent research workflow, presented at the enterprise AI Summit
Prototyped AI-to-AI commerce that transacts autonomously, showcasing the potential of agentic economic systems
Built a tokenized data flow proof-of-concept on testnet using the Ethereum Attestation Service
Architected and delivered batch processing for FINRA datasets, generating $1.5M in business impact

AWS Bedrock
LangGraph
MCP
A2A
AP2
Python
TypeScript
ECS
S3
EAS

$1.5M impact, one pipeline

May 2023 – Aug 2023

Cloud Engineering IT Co-op

Vanguard · Malvern, PA

Architected and built an ECS Insights dashboard to view application metrics through a cost lens
Optimized an existing web service to increase performance by reducing complexity
Analyzed cluster data to build models identifying trends for resource optimization
Synthesized executive summaries presented in department reviews to drive resource efficiency

React
Next.js
TypeScript
Python
Java
AWS
Splunk
Tailwind CSS

the return offer worked

May 2022 – Jan 2023

Cloud Software Engineer Co-op

LenelS2 (Carrier) · Pittsford, NY

Spearheaded mobile development for iOS and Android to establish the structure for native development
Secured billing systems end-to-end to mitigate risk exposure and reduce vulnerabilities
Collaboratively developed a navigation overhaul to enhance user efficiency and accessibility
Standardized and refactored UI components to improve overall user experience

React
Node.js
Swift
Android
Azure
Jenkins
Cucumber

security is a habit now

Oct 2020 – May 2024

Senior Software Engineer, Project Manager & Researcher

Accessible Learning Labs, RIT (NSF-funded) · Rochester, NY

Managed a distributed research and development team across universities
Designed and developed an Ethics of AI learning lab to teach the importance of equitable software
Published and presented a paper on equity and user bias at ICSE 2023
Researched software accessibility and user experience to emphasize equity and inclusivity in design

React
Node.js
PostgreSQL
Tailwind CSS
Figma
Agile

see: evidence ↓

+ B.S. Software Engineering, minor in Business Administration — RIT, class of 2024

toolchain — what i reach for

Agents at work. Agent tools at home.

The through-line isn't a framework — it's pushing whatever AI can do today slightly past where it's supposed to stop.

Agent systems — daily drivers

LangGraph

AWS Bedrock

MCP · A2A · AP2

EAS / on-chain

AWS ECS · S3

…and the boring-but-load-bearing rest: React, Node.js, PostgreSQL, distributed systems, pipelines that actually compile.

Languages spoken

Python
TypeScript
Go
Java
Solidity

// Go entered the chat in 2026 — every CLI since is Go.

AWS Certified AI Practitioner
AWS Certified Cloud Practitioner

Off the clock

Bouldering, lifting, making music. Same appetite, different walls. [unconfirmed — mark's old site]

evidence — peer-reviewed, not vibes

Before I built agents, I studied humans.

“Do Users Act Equitably? Understanding User Bias Through a Large In-Person Study”
Four years on RIT's NSF-funded Accessible Learning Labs — running a research team across universities, building an Ethics of AI lab, and measuring how real people treat software inequitably. It taught me the thing I still apply to agents: systems inherit the biases of whoever they serve, unless you design against it.
sternefeld et al.
ACM/IEEE ICSE 2023
NSF-funded
in-person study @ Imagine RIT
Read on IEEE Xplore— opens in a new tab Accessible Learning Labs— opens in a new tab

Accessibility isn't an afterthought on this site — it's the research background. This page is built to WCAG 2.2 AA.

artifacts — public, inspectable, real

Built because I couldn't not.

Nobody assigned these. That's the point.

triton-cliGo · CDP

A browser for agents. No Node, no Playwright — one binary that snapshots any page as a compact accessibility tree, so an LLM can browse the web for hundreds of tokens instead of tens of thousands.

AES-256 sessions · interactive REPL · zero deps

drydockGo · Claude Code

Say "make me a deck" — get a real .pptx. A Claude Code plugin with a hardened Go orchestrator driving a sandboxed compiler: SSRF guards, zip-bomb checks, XML-injection defense.

natural language → PowerPoint

eas-file-attestationSolidity

Prove a file existed without showing anyone the file — Merkle roots on-chain via Ethereum Attestation Service. The public twin of my work proof-of-concept.

privacy-preserving receipts

tee-time-optimizerlive · React 19 · Grok

A golf pairing engine styled like a trading terminal — weighted compatibility index, LLM-generated matchup "prop bets," CSV in and out. Proof I also ship things with faces.

open app

tavern-management-gamefor fun · his most-starred repo

A medieval bartending sim — pour to the line, keep the queue happy. Try it on the glass, or

play the real one

You've met me. Your move.

Agentic systems, developer tooling, or something that doesn't exist yet — I answer email from humans and well-behaved agents.

initiate handshake

Opens your mail client → markfsternefeld@gmail.com · latency: usually under 24h

github/Mster115 LinkedIn Download résumé (PDF)

Mark Ferenc Sternefeld · every fact above shipped, published, or compiled from source.

Agents do my busywork.*