Tarık Deveci
online

AI Product Engineer İzmir · open to İstanbul

I build systems that perceive, decide, and ship.

AI product engineer who takes ideas to production: LLM pipelines, decision engines, full-stack products and mobile apps, shipped in real domains from municipal AI to industrial safety. I've built a lot on my own — now I want to do it inside a strong engineering team, at real scale, with people I can learn from.

View selected work Download CV GitHub LinkedIn

Ask me about , , or .

scroll
Portrait of Tarık Deveci, AI Product Engineer.

About

I'm Tarık — an AI product engineer who likes carrying an idea all the way: through the model, the backend, the interface and the deploy, until it's live and someone is using it. Fresh out of Software Engineering, I've shipped a municipal AI call-center, a productivity OS, an explainable decision engine, a process-safety platform and my SAM3 thesis. Building those on my own taught me to own a problem end to end; working on Hepsijet's backend team showed me how much further it goes with real engineering process — code review, CI/CD, people who've done it at scale. That's what I want next: a strong team where I can keep shipping and grow. Alongside the AI work, I bring real domain depth in sustainability and climate tech — from carbon-accounting platforms to a formal ESG standards program (ISO 14064, GHG Scope 3). Based in İzmir, open to İstanbul and remote.

  • İzmir, TR
  • Software Engineering grad
  • Sustainability & climate tech
  • 7 shipped · 6 live
  • Open to İstanbul & remote

Selected work

Four shipped products, each framed by the problem, the system built, and the measurable outcome.

  1. preventa · HAZOP workspace
    PreventA HAZOP workspace: a deviation worksheet for a reactor feed pump with cause and consequence rows, the study node tree, and a grounded AI-suggestions panel.
    01Process Safety · Hybrid RAG

    PreventA

    A process-safety platform that turns HAZOP / LOPA methodology into structured, auditable risk software, with a hybrid-RAG layer grounded in plant documentation.

    Role Designer & Engineer Year 2025 Stack FastAPI · PostgreSQL · pgvector · React · Hybrid RAG
    • Structured HAZOP / LOPA worksheets
    • Grounded suggestions, with citations
    Live app Case study
  2. lifeos.tr / dashboard
    LifeOS in-app dashboard: time blocks, AI-ranked tasks, workout, nutrition and weekly rhythm in one synced operating screen.
    02AI Productivity · Mobile + Web

    LifeOS

    A personal operating system for tasks, time blocking, nutrition and workouts, with AI daily planning synced across web and mobile.

    Role Founder & Engineer Year 2025 Stack Next.js · React Native · Expo · Supabase · Zustand · RevenueCat · iyzico
    • 2 platforms, one sync layer
    • 7 AI features
    • Subscriptions live
    Live site Case study
  3. octostra.app / focus
    Octostra dashboard: the Today's Focus engine ranking five tasks by priority score, each with reason codes and a written rationale.
    03Decision Intelligence

    Octostra

    A decision-intelligence engine that tells you what to focus on now, and shows its reasoning. Not another task manager.

    Role Designer & Engineer Year 2025 Stack FastAPI · PostgreSQL · React · Docker
    • Explainable, no LLM black box
    • Top 3–5 with reasoning
    Live app Source Case study
  4. debateai-product / new analysis
    DebateAI product screen: a filled market-entry brief form with company, competitor, industry and target market inputs.
    04Multi-Agent Strategy

    DebateAI

    A market-entry strategy engine where 100+ specialized agents debate options, red-team assumptions, score surviving plans and assemble a board-ready playbook.

    Role Product Engineer Year 2025 Stack Next.js · LLM pipeline · Multi-agent orchestration · Three.js
    • Structured strategy intake
    • Visible agent workflow
    Live product Case study Example run

More projects

Autonomous multi-agent consulting, municipal AI, sustainability, embodied AI, and open source.

Detay İnovasyon site: a sustainability consultancy home with experience, completed-project and CO₂-reduction figures.

Detay İnovasyon

Sustainability

A carbon-footprint tracking platform and digital presence for an environmental consultancy, processing 1,000+ emission records every month.

Live site

React · Node.js

MuniGo

Municipal AI

An AI municipal call-center and CRM, co-founded and built end to end: it routed citizen complaints, tracked cases, and reported KPIs for two real municipalities across 300+ complaint categories.

Reworking the request path and model calls cut spoken-response latency from ~7s to ~1.5s and TTS cost by ~80%.

Case study

Django · PostgreSQL · GPT API · AWS Polly

Ovri

Embodied AI

A small physical companion robot paired with a local-only reasoning core that builds memory and personality over time, designed to integrate with LifeOS and Octostra.

Status: product design and early prototyping. Next up — an on-device memory graph and a companion app that plugs into LifeOS and Octostra.

Local LLM · Memory engine · Robotics

Open source

GitHub

Octostra, Carbonfootprint, TicketingSystem and more, public on GitHub.

Browse repositories

The path

Where I studied, and the roles that each left a measurable mark.

  1. 2026

    BSc Software Engineering · Bahçeşehir University

    Graduated 2026. Capstone: DetectIQ, a SAM3 + LoRA fine-tune for structural damage segmentation, taken to production.

    3.2 / 4.0 GPA
  2. 2025

    Software Engineering Intern · Hepsijet

    My first taste of a real engineering org: backend on production microservices with code review and CI/CD — Java Spring Boot, Docker, Jenkins, Datadog.

  3. 2024–25

    Co-Founder & Lead Software Developer · MuniGo

    Built an AI municipal CRM end to end: Django, PostgreSQL, GPT API, AWS Polly.

    −80% TTS cost · 7s → 1.5s latency · 2 municipalities
  4. 2024

    Junior Full-Stack Developer · AtıkNakit

    Web and mobile delivery with Node.js, PostgreSQL and Flutter.

    ~25% API response improvement
  5. 2022–23

    Software Developer · Detay İnovasyon

    Carbon-footprint tracking platform and web presence with React and Node.js.

    1,000+ monthly emission records
RESEARCH · SHIPPED TO PRODUCTION

DetectIQ: a SAM3 fine-tune for structural damage, from thesis to a live product.

My capstone fine-tunes Meta's SAM3 segmentation foundation model with LoRA to segment bridge and infrastructure damage — cracks, corrosion, spalling and 16 other classes — at the pixel level, on the DACL10K dataset. I trained just 0.25% of its 842M parameters, then shipped the model behind a FastAPI + React inspection platform: upload a photo, get back a colored damage mask with per-class coverage.

0.56mean validation IoU across 19 classes
0.81peak class IoU · 8 of 19 above 0.60
0.25%of 842M params fine-tuned (LoRA)
~10KDACL10K images · 975 validation
  • SAM3 base, LoRA adapters on the attention projections (q_proj / v_proj), ~8MB adapter
  • 10 epochs on an A100 with a combined BCE + Dice loss; best validation loss 0.27
  • Served through a swappable detector seam — mock or the real model via a one-line config flag
Live product Case study Model repo

Tech stack

What I reach for, grouped by where it lives.

Product & Web

  • Next.js
  • React
  • TypeScript
  • React Native · Expo
  • Tailwind
  • Zustand

Backend & Data

  • FastAPI
  • Django
  • Java Spring Boot
  • Node.js
  • PostgreSQL
  • Supabase

AI & ML

  • LLM pipelines
  • Hybrid RAG
  • LoRA / PEFT fine-tuning
  • SAM3 / segmentation
  • GPT API
  • Decision engines

Infra & Ops

  • Docker
  • Jenkins
  • Datadog
  • AWS
  • Vercel
  • RevenueCat · iyzico

open to AI / software engineering roles where I can build and grow

Ready to build inside a strong team.

Best fit: AI Product Engineer, Full-Stack AI Engineer, or product-minded software engineer roles on a team where I can own real problems, learn the process, and grow alongside senior engineers.

Ask about my work

Answers come from an assistant grounded in my real work. Email me for anything else.