You will get a production AI agent or RAG system, built end-to-end
Project details
I'm a Principal Engineer with 12+ years building AI-powered SaaS at scale. 3 years at Apple (London and California), then founded Setzzy — an AI and real-time SaaS that I scaled to $1.45M in funding across 20+ countries and exited in Q1 2026.
I build production AI agents and RAG systems end-to-end. Not a notebook demo, not a single-prompt wrapper — a real system with retrieval, tool calling, evaluation, observability, and a runbook your team can operate after handoff.
Stack: LangChain, LangGraph, AutoGen, CrewAI, Vercel AI SDK, MCP. OpenAI, Anthropic Claude, Google Gemini, and open-source LLMs with model routing. RAG over pgvector or Pinecone with hybrid sparse plus dense search and custom re-ranking. Deployed API on your AWS or cloud, with PostHog, Sentry, and OpenTelemetry hooks.
What sets this apart: Setzzy's matching engine combined OpenAI plus Anthropic Claude with RAG over pgvector and multi-step agent flows in production for 20+ countries. The architecture choices and tradeoffs come from running it, not reading about it.
Best fit: founders and teams who need a real AI agent or RAG system shipped in 3 to 6 weeks.
I build production AI agents and RAG systems end-to-end. Not a notebook demo, not a single-prompt wrapper — a real system with retrieval, tool calling, evaluation, observability, and a runbook your team can operate after handoff.
Stack: LangChain, LangGraph, AutoGen, CrewAI, Vercel AI SDK, MCP. OpenAI, Anthropic Claude, Google Gemini, and open-source LLMs with model routing. RAG over pgvector or Pinecone with hybrid sparse plus dense search and custom re-ranking. Deployed API on your AWS or cloud, with PostHog, Sentry, and OpenTelemetry hooks.
What sets this apart: Setzzy's matching engine combined OpenAI plus Anthropic Claude with RAG over pgvector and multi-step agent flows in production for 20+ countries. The architecture choices and tradeoffs come from running it, not reading about it.
Best fit: founders and teams who need a real AI agent or RAG system shipped in 3 to 6 weeks.
AI Algorithms
Large Language Model, Multimodal Large Language Model, Transformer ModelAI Applications
AI Chatbot, Conversational AI, Natural Language UnderstandingAI Models
ChatGPT, GPT-4What's included
| Service Tiers |
Starter
$4,500
|
Standard
$8,000
|
Advanced
$12,000
|
|---|---|---|---|
| Delivery Time | 21 days | 30 days | 45 days |
Number of Revisions | 1 | 2 | 3 |
AI Model Integration | |||
Batch Normalization | - | - | - |
Database Integration | - | - | - |
Detailed Code Comments | - | - | - |
Image Upscaling | - | - | - |
MLOps | - | - | |
Model Deployment | - | ||
Model Documentation | |||
Model Monitoring | - | - | |
Model Testing & Optimization | - | ||
Model Tuning | - | ||
Natural Language Processing | - | - | - |
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | |||
Setup File | - | - | - |
Source Code | - | - | - |
Frequently asked questions
About Yunus
Senior Full-Stack, AI & Mobile Engineer | Ex-Apple | Next.js, LLMs
Izmir, Turkey - 12:28 am local time
What I build
— AI and LLM Systems
Production RAG pipelines with pgvector, Pinecone, and custom re-rankers. Multi-agent workflows with LangChain, LangGraph, AutoGen, CrewAI. OpenAI, Anthropic Claude, Google Gemini, and open-source LLMs. Voice AI pipelines using Whisper, OpenAI Realtime, and ElevenLabs. Agent evaluation, prompt engineering, model routing, cost optimization, and production observability.
— Full-Stack Web
Next js (App Router), React, TypeScript, Tailwind, shadcn ui, tRPC. Node js, Python, FastAPI, NestJS. REST, GraphQL, gRPC. PostgreSQL with partitioning and replication, Redis, MongoDB, pgvector. Multi-tenant SaaS, Stripe Connect, and full internationalization.
— Mobile
Native iOS using Swift, SwiftUI, UIKit, Combine, and Core Data, plus React Native for cross-platform delivery.
— Real-Time and Voice
WebRTC, Socket io, sub-second turn-taking across cross-border sessions. Kafka, Redis Pub/Sub, SQS, BullMQ, and Temporal for async job orchestration.
— Cloud and DevOps
AWS (ECS, RDS, Lambda, S3, Bedrock), Kubernetes, Docker, Terraform, and GitHub Actions. Sustained 99.9% uptime under thousands of concurrent users at Setzzy.
Selected impact
— Setzzy (Founder and CEO): Built 0 to 1, raised $1.45M, scaled to 20+ countries, 2,800+ early-access signups, strategic exit Q1 2026. Reduced expert sourcing from days to under 60 seconds with a proprietary AI matching engine. Cut time-to-payment by 70% with a milestone-based Stripe Connect escrow.
— Apple (Software Development Team Manager, California): Reduced deployment errors 60% and accelerated release cycles 40% across teams serving 500M+ users. Cut system downtime 25% via ML-based telemetry.
— Apple (Engineering Team Lead, London): Cut API latency 45%. Reduced legacy migration risk 60% through phased monolith-to-microservices transition.
— Independent consulting (2013-2019): Delivered 40+ SaaS, automation, and e-commerce platforms for Adidas, Puma, Trendyol, Logitech, Sabanci Holding, MNG Holding, Fenty Beauty, Anastasia Beverly Hills, iRobot, Alo Yoga, Abbott, Sheraton, ELLE, ColourPop, Pernod Ricard. 90%+ client retention over 7 years.
How I work
Direct, async, ownership-first. I treat your codebase like my own. Comfortable with ambiguous briefs. I translate business goals into shipped product, engineering standards, and reliable delivery cycles. Available immediately, 30+ hours per week, remote globally.
Let's talk if you need
— An AI feature or product built end-to-end, not just a prototype
— A senior engineer who can architect and ship, not just write code
— A technical advisor or fractional CTO for early-stage product decisions
— A founder-engineer who has been on your side of the table
Quick to respond. Happy to hop on a short call to scope things out.
Steps for completing your project
After purchasing the project, send requirements so Yunus can start the project.
Delivery time starts when Yunus receives requirements from you.
Yunus works on your project following the steps below.
Revisions may occur after the delivery date.
Kickoff and architecture decision
30-min call to align on use case, model and framework choice (LangChain or LangGraph, single-agent or multi-agent), vector store, and evaluation strategy. Short architecture doc you approve before code starts.
Ingestion, RAG pipeline, and first responses
Document ingestion with chunking tuned to your content. Vector store setup (pgvector or Pinecone) with custom re-ranking. First end-to-end query loop running by day 5-7 so you can test live.