✦ andreesher
Available for AI roles ✦

✦ Applied AI Engineer

Open to AI enablement roles · Denver, CO

Applied AI Engineer

I put AI to work inside real organizations: agents, RAG assistants, and the workflow automation that makes them stick.

</ code. deploy. enable. />

See the flagship agent Résumé GitHub

press ⌘K to navigate

  • 6+AI systems shipped
  • 3,000+employees served
  • 5 yrstechnical operations
  • 25+technologies

What
I Do ✦

I design, build and ship AI systems that survive contact with real organizations: grounded, measured, and adopted by the people who use them.

AI Agents

Multi-step agentic pipelines: LangGraph orchestration, routing, guardrails and human-in-the-loop handoff.

01

RAG Systems

Retrieval that cites its sources: FAISS, Bedrock and permission-aware pipelines grounded in real data.

02

Workflow Automation

Instrumented automation with per-stage latency, error rates and time-saved: ROI you can point at.

03

Cloud Architecture

AWS & Azure foundations: serverless APIs, VPC networking, and the middleware that makes AI production-safe.

04

flagship agent

Customer Support
Email Agent

A 9-node LangGraph pipeline that reads, classifies, retrieves and replies in seconds, and knows when to hand off to a human.

  1. ingest_email parse sender, subject, thread context
  2. classify_intent intent: refund_request · 0.94
  3. route no complaint flags → automated path
  4. retrieve_kb FAISS top-3 chunks · sim 0.89
  5. check_confidence 0.91 ≥ 0.75 → proceed
  6. generate_response GPT-4 drafts grounded reply
  7. quality_gate tone ✓ · grounding ✓ · PII ✓
  8. send_reply sent in 2.9 s end-to-end, was ~30 min
  9. human_review complaints & low confidence never auto-send

FastAPI · LangGraph · FAISS · GPT-4 · SQLAlchemy · 6 decoupled services

Repository ↗

Featured
Projects

View all projects →
02

Marketing Workflow Automation

Multi-stage agentic pipeline: Claude classifies, urgency-scores and routes leads, then drafts personalized follow-ups. Instrumented per-stage latency, error rates and time-saved for ROI measurement.

Python · Claude · Flask · SQLite · Instrumentation

Repo ↗
03

AWS Serverless Task API

Full-stack serverless task scheduler: React/TypeScript front end on an API Gateway → Lambda → DynamoDB back end. Debugged cross-service IAM failures to production stability.

API Gateway · Lambda · DynamoDB · React · TypeScript

Repo ↗
04

Gmail AI Assistant Agent

Autonomous Perceive → Decide → Act agent: Claude summarizes, categorizes and urgency-scores incoming mail, then replies through the Gmail API with OAuth 2.0.

Python · Gmail API · OAuth 2.0 · Claude

Repo ↗
05

Denver Broncos NFL IQ

RAG system grounding Claude 3 Haiku (AWS Bedrock) in 17,000+ rows of NFL statistics, so every answer cites real data instead of hallucinating.

AWS Bedrock · Claude 3 Haiku · boto3 · pandas · RAG

Repo ↗
06

Three-Tier AWS Web App

Production-grade network from scratch: custom VPC, 8 subnets across 2 AZs, dual ALBs, auto-scaling Flask tier (2-4 instances), encrypted RDS Postgres in isolated subnets.

VPC · EC2 · ALB · Auto Scaling · RDS · Nginx · Gunicorn

Repo ↗

$ gh api users/Aesher7

github.com/Aesher7 ↗

I ran the operations.
Now I build the automation.

Five years inside fleet-scale technical operations taught me exactly where manual work breaks. Then I learned to build the systems that fix it.

  1. 2020 to 2026

    Sunrun

    Service technical efficiency

    Telemetry, log analysis and root-cause work across thousands of distributed solar systems. Remotely remediated 25% of escalated cases, cutting field-dispatch cost.

  2. 2024

    Qookit AI

    Software engineer intern

    First applied-AI shipping: modular JavaScript front ends and Flutter apps wired to Python AI pipelines, at startup pace.

  3. 2024 to 2027

    CSU Global

    B.S. CS → M.S. AI & ML

    3.82 GPA, concentrations in artificial intelligence and machine learning, the theory behind the systems I ship.

  4. 2026 to now

    IMA Financial

    Applied AI Fellow

    Built and deployed an HR knowledge assistant serving 3,000+ employees (Copilot Studio in Teams, grounded with citations), and led RAG-architecture discovery across HR, IT, security and AI-governance stakeholders: Azure Functions middleware, Key Vault, permission-aware retrieval.

Tools I reach for

AI & automation

  • LLMs
  • RAG
  • AI agents
  • LangGraph
  • Prompt engineering
  • Permission-aware retrieval
  • Identity mapping
  • Workflow automation
  • Fine-tuning
  • Instrumentation

Enablement platforms

  • Microsoft Copilot Studio
  • Teams
  • Entra ID
  • Azure Functions
  • Azure Key Vault
  • Azure OpenAI
  • Power BI
  • Power Automate
  • Salesforce CRM

AWS cloud

  • Lambda
  • API Gateway
  • Bedrock
  • EC2
  • S3
  • DynamoDB
  • RDS
  • VPC
  • IAM

Languages & frameworks

  • Python
  • JavaScript / TypeScript
  • Dart
  • React
  • Flutter
  • Docker
  • Git
  • pandas
  • boto3

Education

B.S. Computer Science → M.S. Artificial Intelligence & Machine Learning

Colorado State University Global

GPA 3.82 / 4.00

Certification

AWS Certified Cloud Practitioner CLF-C02

Earned March 2026

Core services, architecture, pricing and security

Verify on Credly ↗

AI, put to work
down the street.

I help small businesses around Berthoud and Northern Colorado adopt AI that actually earns its keep, with no in-house tech team required. Practical tools, set up to fit how you already run things.

What that looks like

  • A real website, built and shipped fast.
  • Automating the repetitive work: inbox, scheduling, follow-ups.
  • AI tools chosen to fit your day, not a rebuild of it.
  • Straight answers on what's worth using, and what isn't.
Proof

Solana Dog Grooming

Craig, CO · Pet grooming

Built their business website end to end with an AI-assisted prompt-chaining workflow. Design, copy and build in one pass, for a real local shop.

AI-assisted build · prompt chaining

Call 720-753-2417 Email me Berthoud, CO · serving Northern Colorado

Let’s build
something.

andreesher1@gmail.com