Independent AI consultant & builder

I build AI that ships — and train the teams that run it.

I'm Kamran Darugar. I help companies design, build, and deploy production AI systems — and get their people fluent enough to own them. From a loose idea to a system in production, end to end.

Start a project→See services→

Founder of GutChat·Production AI across legal, biotech, finance & telecom

Portrait of Kamran Darugar
San Diego—:—

Trusted to build for

  • Superagentic AI
  • Veterans United Home Loans
  • WAN-AI
  • LegalEase Solutions
  • WinSanTor
  • SDSU

What I do

Two ways to work together.

01

Build

Custom AI & software solutions

I design and ship the system — not a slide deck.

End-to-end engineering for teams that need a real AI product in production: assistants, agents, RAG pipelines, and the infrastructure underneath them. I take the work from a loose idea to a deployed, observable system your team can run.

  • —Discovery, scoping & technical architecture
  • —LLM, RAG & agentic systems (OpenAI, Bedrock, custom)
  • —Full-stack build — backend, data, and the interface on top
  • —Scalable, secure cloud infra on AWS (load-balanced, auto-scaling)
Start a project→Details→

02

Train

AI proficiency for your team

Your people fluent in AI — not just impressed by it.

Hands-on training and enablement that turns AI from a buzzword into a daily tool. I run practical workshops grounded in how these systems actually behave in production — what they're good at, where they break, and how to use them responsibly.

  • —Practical, hands-on workshops tailored to your team's work
  • —Prompting, RAG & agent best-practices from real deployments
  • —Tooling setup & workflow integration for daily use
  • —Custom curriculum mapped to your domain and goals
Book training→Details→

How it works

A clear path from idea to shipped.

  1. 01

    Discover

    We pin down the problem, the constraints, and what success actually looks like. No build starts without a clear target.

  2. 02

    Architect

    I design the system — data, models, infra, and interface — and show you the plan before a line of production code is written.

  3. 03

    Build / Train

    Tight iterations with working software (or live workshops) in your hands early and often, evaluated against real outcomes.

  4. 04

    Ship & hand over

    Deployed, documented, observable. Your team leaves knowing how to run, extend, and trust what we built.

Proof

Systems shipped, outcomes measured.

  • 01

    2025 →

    Product · AI Assistant

    Veterans United Home Loans

    A production AI assistant for veteran.com

    A full-scale assistant on the OpenAI APIs, served through a microservices architecture with a real-time analytics pipeline and a distributed data warehouse behind it — built to handle high-volume user interactions on secure, auto-scaling AWS infrastructure.

    • OpenAI
    • AWS
    • Microservices
    • Real-time analytics

    Live

    Serving veteran.com at scale

  • 02

    2023

    Consulting · Gen-AI

    SDSU Alumni

    Gen-AI gift-reporting assistant

    A Gen-AI assistant for SDSU's Financial Management Gift Reporting team — built on the OpenAI APIs, surfaced through a Django web interface, with a tuned data-query tool that transformed a slow manual reporting workflow.

    • OpenAI
    • Django
    • Data tooling

    3×

    Reporting efficiency gain

  • 03

    2024 – 2025

    Internal tool · Legal AI

    LegalEase Solutions

    NDA review-and-revision pipeline

    An event-driven backend that locates client NDAs in ClickUp, categorizes them in AWS, queues them to Bedrock for parallel automated review, and ships an internal UI behind a horizontally-scaled, containerized deploy.

    • AWS Bedrock
    • ClickUp
    • Docker
    • Event-driven

    Parallel

    Automated review at scale

  • 04

    2022 – 2024

    Product · Biotech ML

    WinSanTor

    ML automation for biopsy imaging

    A Django interface over a machine-learning program that automates biopsy image markings, with region- and data-extraction scripts for medical image formats and a Dockerized PyTorch inference engine wired in for interactive use.

    • PyTorch
    • Django
    • Docker
    • Medical imaging

    +200%

    Data-extraction throughput

All case studies→

Venture

Founder

GutChat

Turn gut feelings into gut facts.

An AI health app I founded and built end-to-end. It turns daily logging — meal photos, sample scans, weight-from-photo, sleep, stress, and water — into a composite daily gut score, with personalized coaching that remembers your history, experiments that surface your triggers, and doctor-ready reports.

  • Computer vision for meal & sample analysis
  • Composite daily gut-score model
  • Coaching with long-term memory
Visit gutchat.org↗

Writing

Field notes on building with AI in production.

Get started

Have an AI project worth building together?

kamrandarugar@gmail.com

Replies usually within a day