KW
Open to Opportunities

Kevin Welty

Senior Software Engineer · AI & Agentic Systems

10+ years building backend systems, ETL pipelines, and agentic AI workflows

Accomplished Senior Software Engineer with over 10 years of experience designing, developing, and deploying robust backend systems and data pipelines. Currently specializing in LLM-powered agentic workflows and intelligent automation — bridging deep infrastructure expertise with cutting-edge AI tooling to ship production-grade AI systems.

Skills & Technologies

Languages

PythonJavaC#GoRubyLuaSQL

AI & Agentic

Agentic WorkflowsAnthropic ClaudeClaude APILarge Language ModelsRAGPrompt Engineering

Frameworks & Tools

FastAPIDjangoRay/AnyscaleSQLAlchemyAlembicPydanticJinjaXSLT

Databases

PostgreSQLMS SQL ServerCassandraMongoDBMySQLElasticSearchOpenSearch

Infrastructure & Cloud

AWS (S3, SQS, SNS)Google Cloud PlatformLinuxDockerCI/CD

Protocols & Patterns

RESTful APIsOAuth 2.0JWTMicroservicesSNMPWiresharkUnit TestingRelease Management

Work History

Lex Machina · Remote

October 2021 – Present

Senior Software Engineer II

  • Maintain backend crawling infrastructure spanning multiple ETL pipelines that process millions of legal documents.
  • Contributed to projects adding new state courts and millions of new documents to the product.
  • Optimized the federal crawling backend, improving reliability and data freshness.
  • Migrated several crawling systems to a new infrastructure stack with zero downtime.
  • Designed database schemas with Alembic migrations and SQLAlchemy models.
  • Engineered event-driven data ingestion pipelines on AWS S3, SQS, and SNS.
  • Built high-performance RESTful APIs with FastAPI — Pydantic validation, dependency injection, and async processing.
  • Developed secure API integrations using OAuth 2.0, JWT, and API keys for Google APIs and custom endpoints.
  • Created proofs of concept for new solutions; active in the Python community (3 PyCons, including onsite in 2024).

Comtech — EF Data & Satellite Networks Technologies · Tempe, AZ

March 2016 – October 2021

Software Engineer I

September 2019 – October 2021
  • Developed a Python client/server automated test application to validate the Heights Networking System.
  • Transformed a beta Python application (SMT) into microservices: result collection, dynamic views, config saving, and a RESTful interface.
  • Created a RESTful API framework on a Django web server serving test results, rack configs, and unit configurations.
  • Served as lead engineer and developer/tester across multiple simultaneous projects with major architectural decisions.

Applications Software Engineer

May 2019 – September 2019
  • Lead engineer for Skyline NetVue Network Management System, supporting networks for commercial and government clients.
  • Worked directly with customers to troubleshoot field issues; debugged satellite networks at the packet level with Wireshark.
  • Developed a DataMiner REST API test application enabling customers to beta test API calls.
  • Initiated development of an in-house NMS to replace NetVue, leading a team of four.

Network Test Engineer I & II

September 2016 – May 2019
  • Administered and developed the Java Test Automation Application (Test Harness) using SQL, Java, Python, and C++.
  • Developed Wireshark dissectors in Lua for human-readable packet analysis.
  • Built packets with Scapy in Python and tested their impact on satellite network systems.

Engineering Intern

March 2016 – September 2016
  • Developed equipment drivers using REST, GPIB, SNMP, and Selenium.
  • Built automation software and test procedures for satellite RF and network communications equipment.
  • Developed a C# Telnet debugging application for customer field use.

Selected Projects

Projects spanning production backend systems, agentic AI workflows, and data infrastructure.

Agentic Backend Data Pipeline

shipped

An agent-driven pipeline where Claude builds the backend systems that collect and decompose data for analytics — no hand-coding each solution. Multiple workstreams run in parallel while I direct the agent and validate outcomes instead of writing boilerplate.

PythonFastAPIClaude APIPostgreSQLETL

AI Coding Standards Enforcement

shipped

A system of custom skills and structured prompts that teaches AI models company-specific coding standards: naming conventions, architectural patterns, and file structure. AI-generated code matches the existing codebase from the first output, cutting review back-and-forth.

Claude APIPythonPrompt EngineeringCI/CD

Claude-Powered Anomaly Detection & Alerting

shipped

Claude integrated into production monitoring as an intelligent layer beyond static thresholds. It analyzes system output, logs, and data patterns, then fires alerts with a human-readable explanation of what went wrong and why it matters — context instead of raw noise.

Claude APIPythonFastAPIAlertingObservability

What I'm Working On

Current areas of focus — where I'm spending cycles right now.

Agentic Workflows in Everyday Work

Actively embedding agentic AI into daily work routines and personal projects — not just one-off experiments, but finding where agents can reliably reduce friction, handle repetitive tasks autonomously, and free up mental bandwidth for work that actually requires human judgment.

AI Enablement for Small Businesses

Sharing AI knowledge with individuals and small business owners who want to boost productivity but don't have deep technical backgrounds. Helping them cut through the noise, identify where AI genuinely saves time, and get hands-on with tools that make an immediate difference.

Agentic Apps for Backend Productivity

Building applications that use agentic AI to take over backend work autonomously — letting the agent handle implementation while I focus on architecture, validation, and outcomes. The goal is multiplying what a single engineer can ship in a given sprint.

Coding Standards for Claude

Developing and refining structured coding standards that Claude can internalize — naming conventions, architectural patterns, file organization, and language-specific rules. The result is AI-generated code that fits the existing codebase from the first output, without a round of cleanup.

Self-Correcting Agent Loops

Experimenting with agent loops that observe their own failures and course-correct without human intervention. When a step breaks, the agent diagnoses, re-routes, and retries on its own — pushing toward truly autonomous execution on complex, multi-step tasks.

BS in Computer Information Systems

Arizona State University

Tempe, AZ · January 2016

Eagle Scout

The highest rank achievable in Boy Scouts of America — earned by fewer than 4% of scouts through demonstrated leadership, community service, and merit.

Let's Work Together

Whether you're building agentic AI systems, modernizing backend infrastructure, or need a senior engineer who can move fast and ship — I'd love to talk.

kjwelty0916@gmail.com
GitHubLinkedInSan Tan Valley, AZ