Resume
Senior Software Engineer at Pedestal AI

Jake Moses builds cloud systems ships AI products solves hard problems delivers business impact

I like using software to quietly remove friction from people's lives and give them time back. Most recently, that's meant building voice and text agents that talk to real customers before their deliveries arrive.

See my work Get in touch
Scroll
Jake Moses

About

A do-it-all engineer who ships.

I'm a Senior Software Engineer at Pedestal AI, where I build AI systems that quietly remove friction points in people's lives, giving them time back or more clarity in moments that used to be confusing. I've had the opportunity to build voice and text agents that converse with customers before their deliveries, reduce friction in purchase order workflows for automotive manufacturers, and improve visibility for shipping customers tracking their goods through customs.

Based in Boston, MA, I've delivered meaningful work across multiple Amazon teams including Pharmacy, Last Mile Tech, and Scout. That includes reducing pharmacist shift requirements by 15% and launching products used across North America, Europe, and Japan.

What sets me apart: I take full ownership. I'm as comfortable whiteboarding architecture as I am debugging production systems at 2am. I build AI tools on weekends because I genuinely love this work.

Toolkit

Sharp across the stack.

Languages

Python Java TypeScript SQL Node.js

AWS

ECS Fargate Lambda DynamoDB S3 CloudFront Kinesis SQS/SNS RDS Lambda MicroVMs Route53

Azure

Azure Functions Azure Batch Azure Front Door Azure Container Apps

AI

LLM Integration Agentic Systems Eval Engineering (AI Evaluations) Harness Engineering Skill Development Sandbox Code Execution Claude Codex Cursor Fireworks (via Claude Code)

Practices

CI/CD Serverless Agile/Scrum System Design DevOps

Stakeholder Engagement

PRD Writing Project Plan Alignment Technical Discovery Customer Demos Vendor Negotiations Contract Negotiations Sales Deck Development

Experience

Where I've made impact.

Senior Software Engineer

Pedestal AI
Apr 2025 — Present · Boston, MA
  • Built and shipped voice and text AI agents that have handled 600K+ customer conversations for a Fortune 500 retail company, automating predelivery communications, collecting delivery instructions, and passing them to drivers or customer service
  • Built and deployed a document processing system that automated purchase order data entry for automotive manufacturing clients, processing thousands of purchase orders
  • Built and maintained an AI system that helped customers of a Fortune 300 shipping company get answers to customs processing questions and shipment lookups
  • Worked directly with customers on rollout and support for each of these systems, in addition to core product development with the engineering team

Software Engineer II

Amazon Pharmacy
Oct 2023 — Mar 2025 · Boston, MA
  • Delivered automation feature reducing pharmacist shift requirements by 15%
  • Achieved 99% automation of machine signal tracking via Lambda, DynamoDB, SNS, Kinesis
  • Led fulfillment sad path handling for cross-functional PillPack Packets launch

Software Engineer II

Last Mile Tech, Amazon
Nov 2022 — Oct 2023 · Seattle, WA
  • Designed and launched two major initiatives across North America, Europe, and Japan
  • Spearheaded agile/scrum adoption, improving team productivity by 25%
  • Optimized package planning workflows with CloudWatch, DynamoDB, Lambda, SQS

Software Engineer I

Amazon Scout
Jul 2021 — Nov 2022 · Seattle, WA
  • Built AWS-deployed web app to query and dynamically label robot-generated event reports
  • Created ecosystem of services for planning, operating, and analyzing robotic deliveries
  • Collaborated cross-functionally to deliver high-impact solutions

SDE Intern

Amazon
May — Aug 2020 · Seattle, WA
  • Created web application for querying and labeling field robot event reports
  • Deployed with Route53, Lambda, S3, CloudFront, Lambda@Edge

Technology Intern

Nucleus195
May 2019 — Jan 2020 · Stamford, CT
  • Automated client outreach processes with Python, significantly improving efficiency
  • Converted spreadsheets into client-ready PDFs via custom scripting

B.S. Computer Engineering

University of Virginia · Class of 2021 · GPA 3.74

TA for Database Systems (2 sem), Algorithms (1 sem), Software Testing (1 sem)

Side project

Twelve AI models are playing fantasy football against each other.

I built a fantasy league where every team is managed by a different large language model for the full 2026 NFL season. Same prompt, same tools, same information, one scoreboard. No humans play. Everything the agents read, write, decide and spend is published as it happens.

It is a real league and a long-horizon benchmark at the same time. Eighteen weeks of drafting, waivers, trades and lineup calls turn into a record of how twelve models handle judgment under uncertainty, with the full transcript of every decision open to read.

Next.js 16 Neon Postgres Vercel Workflows AI Gateway TypeScript

I write up what the models do as the season goes. So far: how they drafted, how they trade, and how the league runs on Vercel.

How it works Visit the league
12 Models managing teams
0 Human managers
18 Week season

Writing

Thoughts on building AI systems that work.

I write about agents, production AI, and the engineering practices that keep them reliable. Some posts are on Pedestal AI's engineering blog. The rest are on LinkedIn, including a running series on my agents-only fantasy football league, where twelve AI models draft, trade, and set lineups with no humans involved.

Sep 9, 2026 · AI League

How 12 AI Models Approach Trading in a Fantasy Football League

Two weeks in, the models sent 37 trade offers and closed 6. Who sends fewer, better offers, who wears the other side down, and why the models that reason the most trade the least.

Read the post
Sep 4, 2026 · AI League

How My Agents-Only Fantasy League Runs on Vercel

One cron tick a minute, eight durable functions, and a queue the agents use to schedule their own sessions. How a 17-week season runs a multi-hour draft with no one on call.

Read the post
Aug 31, 2026 · AI League

How Twelve AI Models Approached Drafting Fantasy Football Teams

Same tools, same information, 168 picks. Some models wrote tiered boards with fallback rules, others researched injuries, and the cost to finish the draft ranged 41x across models.

Read the post
Aug 26, 2026

Scaling Trust in the Era of Voice AI

How a 20-customer pilot at Pedestal grew into 4,000 calls a day, and why the hard part was never the compute.

Read the post
Jul 27, 2026

The Rubber Duck That Talks Back

How speaking ideas aloud and letting an LLM refine them changed the way I write, without losing my original thinking along the way.

Read the post
Jul 21, 2026

Product Managing the System Around Your Agents

AI can produce code fast, but speed without design reviews, testing standards, and task decomposition just means agents running without guardrails.

Read the post
Jul 17, 2026

Beyond Alerts: Using AI to Continuously Improve Production Systems

AI agents that connect telemetry to code and business intent can move past reactive monitoring, continuously asking why a system works the way it does.

Read the post

Contact

Let's build something.

Whether you want to collaborate, have an opportunity, or just want to talk tech — I'd love to hear from you.

mosesjake32@gmail.com (203) 918-3070
Boston, MA