Upwork home
Sign up
Project Catalog
DesignDesign
Video & AudioVideo & Audio
Development & ITDevelopment & IT
MarketingMarketing
Writing & TranslationWriting & Translation
MoreMore

You will get a full audit of your AI app with a plan to cut LLM costs by 40%

Jaeho B.Status: Offline
Jaeho B. Jaeho B.
Preview image

Let a pro handle the details

Buy Other AI & Machine Learning services from Jaeho, priced and ready to go.
Jaeho B.Status: Offline
Jaeho B. Jaeho B.
Preview image

Let a pro handle the details

Buy Other AI & Machine Learning services from Jaeho, priced and ready to go.

Project details

Are your OpenAI or Claude API bills growing faster than your
revenue? I'll find exactly where you're wasting tokens—and
tell you how to fix it.

I've reduced LLM operational costs by 40-55% across my own
production systems using intelligent model routing, prompt
consolidation, and orchestrator-subagent architecture. I built
LightHarness, a 3-tier agent framework that routes tasks across
Claude Opus, Sonnet, Gemini Flash, and local LLMs based on
complexity—cutting average cost per workflow by 52%.

This audit applies that same methodology to your system.

You'll receive a prioritized written report identifying your
top cost drivers, specific fixes with estimated savings per
item, and a clear implementation roadmap. No fluff, no generic
advice—every recommendation is specific to your codebase and
usage patterns.

Works with any LLM provider: OpenAI, Anthropic, Gemini,
Mistral, or self-hosted models.

NEW TO UPWORK — INTRODUCTORY RATE. Senior-level AI
architecture experience at a starter price. This rate
won't last long.
AI Development Type
Model Tuning, Software Maintenance
AI Tools
MATLAB, MLflow
AI Development Language
Python
What's included
Service Tiers Starter
$149
Standard
$299
Advanced
$599
Delivery Time 3 days 5 days 10 days
Number of Revisions
123
AI Model Integration
-
-
Detailed Code Comments
-
-
Knowledge Graph
-
-
-
Model Documentation
Ontology
-
-
-
Source Code
-
-
Taxonomy
-
-
-
Optional add-ons You can add these on the next page.
Fast Delivery
+$50 - $150
Additional Revision
+$50
Written Q&A Follow-up
+$49

Frequently asked questions

Jaeho B.Status: Offline

About Jaeho

Jaeho B.Status: Offline
AI Automation Engineer | Claude & LLM Integration | Cost Reduction
Seoul, South Korea - 12:16 am local time
Full-Stack AI Engineer | Claude & Gemini Integration | Production SaaS

I build production AI systems that cut LLM operational costs by
40-55% through intelligent multi-agent orchestration. From Claude-powered
SaaS platforms (MaxTrans) to autonomous trading pipelines, I ship
end-to-end solutions—not prototypes.

▸ NEW TO UPWORK — INTRODUCTORY RATE
I've been building AI systems and production SaaS for years, but
this is my first month on Upwork. I'm pricing at $35/hr (well below
my market rate) to build a strong track record here. Early clients
get senior-level work at a starter price—this window won't last long.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

WHAT I DELIVER

▸ AI SaaS MVPs (2-6 weeks from spec to paying users)
▸ Multi-agent orchestration systems with cost-optimized model routing
▸ Custom Claude/Gemini integrations beyond simple API wrappers
▸ Production infrastructure on Railway, Vercel, Cloudflare

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

FEATURED WORK

▸ MaxTrans (Live SaaS) — AI document translation platform with
Stripe subscription + token-based billing, AES-256-GCM encryption,
and tRPC architecture. React 19 + Next.js + PostgreSQL.

▸ LightHarness — Custom 3-tier orchestration framework routing
tasks across Claude Opus → Sonnet → Gemini Flash → local LLM
based on complexity. Reduced average cost per workflow by 52%.

▸ GemWriter — Automated content generation Slack bot using
dual-LLM pipeline with Notion CMS integration. Runs on
GitHub Actions, zero manual intervention.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

TECH STACK

▸ Languages: TypeScript, Python, Node.js
▸ Frontend: Next.js, React 19, Vite, Tailwind
▸ Backend: FastAPI, Express, tRPC
▸ Database: PostgreSQL, MySQL, Drizzle ORM
▸ AI: Claude API, Gemini API, Ollama, MCP, multi-agent systems
▸ Infra: Railway, Vercel, Cloudflare R2, GitHub Actions
▸ Payments/Auth: Stripe (subscription + metering), Better Auth, OAuth 2.0
▸ Security: AES-256-GCM, JWT

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

WHY HIRE ME

✓ Cost-First AI Architecture — I measure tokens the way you
measure dollars. Smart model routing means your OpenAI/Claude
bill stops scaling linearly with usage.

✓ Ship-Ready, Not Demo-Ready — Every project includes
auth, billing, monitoring, and production deployment.
No "works on my machine" handoffs.

✓ Deep AI Specialization — I don't "add ChatGPT" to apps.
I design agent systems that make real decisions: task
decomposition, intelligent routing, retry logic, fallbacks.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

IDEAL PROJECTS

▸ AI SaaS MVP or V2 rewrite ($3K-$15K fixed)
▸ LLM cost audit + refactor for existing production systems
▸ Multi-agent workflow automation (n8n alternative, custom code)
▸ Claude/Gemini integration into existing SaaS

Not a fit: WordPress plugins, no-code-only builds, basic chatbot
wrappers around GPT.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Let's ship something real. Send me your problem—I'll tell you
within 24 hours if I'm the right fit, with a concrete plan.

Steps for completing your project

After purchasing the project, send requirements so Jaeho can start the project.

Delivery time starts when Jaeho receives requirements from you.

Jaeho works on your project following the steps below.

Revisions may occur after the delivery date.

Usage Analysis

Review your API dashboard and architecture to map all LLM call points and identify token consumption patterns.

Cost Driver Identification

Pinpoint top 3-5 inefficiencies — oversized prompts, wrong model selection, redundant calls, missing caching.

Review the work, release payment, and leave feedback to Jaeho.