Skip to content
Qasim KhanBook a call (opens in a new tab)

I turn LLMs intosystems that savetime and money.

Full Stack AI Engineer, 5+ years. Production LLM apps, RAG and MCP integrations, built end to end.

See the workBook a call (opens in a new tab)
  • $1.50 → $0.75per minute, medical translation
  • −80%manual search time
  • −80%proposal writing effort
  • +40%website traffic
  • +20%revenue
  • 18tools in one MCP server
  • −25%API response time
  • −30%server load
Qasim Khan in a brown jacket and round sunglasses, with pine-covered hills behind him.

About

Hi, I'm Qasim.

I build LLM products that run inside real businesses, from healthcare translation to Microsoft 365 agents. I work across the whole stack: model, data, integrations and interface.

  • 5+ years
  • Healthcare and enterprise AI
  • BS Computer Engineering

What I build

What you can hire me for.

  • LLM apps

    Chat products for teams and customers.

    −50%translation cost per minute

  • RAG and memory

    Answers from your own documents, scoped by role.

    Multi-tenantwith role-based access

  • MCP and integrations

    Models that work inside Microsoft 365, NetSuite and more.

    18 toolsin one public MCP server

  • Full-stack delivery

    APIs, interfaces, cloud and CI/CD around the model.

    −25%API response time

How I work

Six steps, every project.

  1. Step 1

    Requirements

    Who it's for and what done means.

  2. Step 2

    PRD

    Scope and success, written down.

  3. Step 3

    Architecture

    Data, services, model calls, trust boundaries.

  4. Step 4

    Implementation

    One phase at a time, typed end to end.

  5. Step 5

    Automated tests

    Tests, plus evals for RAG and agents.

  6. Step 6

    Human verification

    I check it myself before it ships.

Selected work

Systems and what they changed.

  1. 01 / 05

    Real-time medical translation

    A real-time AI translation platform for medical conversations that reduced reliance on human interpreters.

    Translation cost
    $1.50 to $0.75 per minute
    Interpreter dependency
    Reduced
    Harmoni Translate website: an AI medical interpreter app for clinics, with a live Spanish to English transcript card.

    harmonitranslate.com, the public product site.

    How it works

    Simplified, illustrative diagram

    Real-time medical translation

    1. Clinicians run in-person, video-link and phone sessions, review transcripts and SOAP notes from one console.

      • Patient link
      • Patient phone

    6. Translation cost per minute

      from $1.50 to $0.75per minute

  2. 02 / 05

    Lessonly AI

    An AI workspace for teachers: governed chat, agents, specialist MCP tools, essay grading and scheduled project work.

    Features in one workspace
    9
    Specialist MCP services
    6

    How it works

    Simplified, illustrative diagram

    Lessonly AI

    1. Teachers chat, pick agents, launch lesson plans, grade essays and run Cowork projects in one app.

      • Essay grading

    7. 6 specialist MCP services

      9 features in one app

  3. 03 / 05

    Multi-tenant enterprise AI assistant

    ChatGPT-style apps with role-based access, memory and RAG, connected to the systems teams already use.

    Used by
    Internal teams and customers
    Access
    Multi-tenant, role-based

    How it works

    Simplified, illustrative diagram

    Multi-tenant enterprise AI assistant

    1. An employee asks in a ChatGPT-style chat.

    6. Built for

      Many tenants, each with its own roles

  4. 04 / 05

    MCP servers and agent workflows

    Model Context Protocol servers that let LLMs work with OneDrive, Outlook and SharePoint through Microsoft Graph.

    Public on GitHub
    3 servers
    SharePoint server
    18 tools

    How it works

    Simplified, illustrative diagram

    MCP servers and agent workflows

    1. An LLM app or agent decides it needs a tool.

    4. Public on GitHub

      3 Microsoft 365 MCP servers

  5. 05 / 05

    Prompt Injection Arena

    An inspectable demo of why an AI assistant should propose actions, not authorize them.

    Policy evaluation suite
    10 cases
    Scenarios
    3

    How it works

    Simplified, illustrative diagram

    Prompt Injection Arena

    1. Local lexical retrieval pulls in documents, each tagged with its source trust.

    5. In protected mode

      Unverified text alone can't trigger high-risk actions

Toolkit

What I build with.

Contact

Tell me what you're building.

Book a call

Pick a time that works for you.

Pick a time (opens in a new tab)

Or send a message