BARIŞGÜLEÇ

Full-Stack AI Engineer

Applied AI & LLM Systems

Scroll

Building applied AI systems from model pipeline to production.

Full-Stack AI Engineer building production LLM, multi-agent, machine learning, and automation systems across data pipelines, backend, frontend, and deployment. Currently developing AI products at ESTAL in Barcelona. Computer Engineering graduate ranked 2nd in class.

Based in Bursa, Türkiye - working remotely with Barcelona, Spain

Portrait of Barış Güleç

Experience

Apr 2026 - Present

Full-Stack AI Engineer — ESTAL

Barcelona, Spain - Remote

  • •Architected multi-agent AI workflows with Agno and LangGraph, combining tool calling, stateful orchestration, validation gates, and human review behind typed FastAPI services deployed with Docker.
  • •Built a Python, pandas, and scikit-learn costing pipeline with confidence and recency features for explainable EUR/ton and EUR/unit predictions.
  • •Delivered a React/Vite and FastAPI LinkedIn intelligence platform with scheduled enrichment pipelines, transactional MSSQL publishing, and Power BI-ready datasets.
  • •Productionized Playwright invoice automation and PaddleOCR/Gemini document pipelines for structured PDF extraction, anonymization, and asset removal.
Oct 2025 - Present

Founder & AI Product Engineer — Canada Next

Independent

  • •Built and monetized an AI-native media platform, growing its YouTube channel to 24K+ subscribers and 6M+ views through an end-to-end AI production system.
  • •Fine-tuned English TTS models with Unsloth on NVIDIA A100 infrastructure for multi-speaker voice cloning and low-latency speech generation.
  • •Fine-tuned FLUX.1-dev and unified speech, image, research, scripting, video production, and publishing into one AI workflow.
  • •Developed a native SwiftUI news app backed by Supabase and Gemini, aggregating 14 Canadian feeds with scheduled Edge Functions and APNs notifications.
Jul 2025 - Sep 2025

AI Engineer Intern — HAVELSAN

Ankara, Türkiye

  • •Modified and fine-tuned RT-DETR and YOLO architectures, maintaining over 95% detection accuracy while reducing inference latency.
  • •Generated scene-specific synthetic training data with Stable Diffusion and IP-Adapter for rare computer-vision scenarios.
  • •Applied CLAHE-based dehazing and optimized GPT-OSS-20B for specialized inference workloads.
Sep 2024 - Aug 2025

AI Engineer — Fortune Teller Coffee

Remote

  • •Developed an agent-based LLM product with FastAPI, Firebase, Docker, RAG pipelines, and LangChain-based reasoning workflows.
  • •Fine-tuned Gemma-3 27B on NVIDIA A100 infrastructure and integrated the model into a production application.
  • •Owned AI integration and backend architecture for a scalable, user-facing generative experience.
Jul 2024 - Oct 2024

AI Engineer Intern — Dpointgroup

Barcelona, Spain - On-site

  • •Processed and optimized business data with pandas and NumPy, applying VarianceThreshold feature selection and Word2Vec embeddings.
  • •Fine-tuned Llama and Qwen models for task-specific use cases and developed JavaScript automation workflows.

Selected Work

Production AI products spanning mobile, web, multi-agent systems, and computer vision.

2026Featured

SafeDose - AI Medication Safety Platform

Built an App Store mobile application and web platform for AI-powered drug-drug and drug-food interaction analysis. Created a custom SGK-based drug dataset with a pandas ETL pipeline. Placed 2nd at the AI'mpact Full-Stack AI Agent Hackathon with Kapsül AI.

SwiftUIFastAPIGeminipandasSupabase
2026

Multi-Agent n8n Chatbot System

Designed and developed a multi-agent n8n chatbot system using Google Gemini, MongoDB, PostgreSQL, and Docker, supporting dynamic session management, tool calling (function calling), and JWT-based authentication.

n8nGoogle GeminiMongoDBPostgreSQLDockerJWT
2025

HAVELSAN AI Internship: Vision & Generative Pipeline

Delivered a CV/GenAI pipeline including fog removal with Dark Channel Prior, ship image generation via DCGAN, Stable Diffusion 3.5 + IP-Adapter image-to-image enhancement, and RT-DETR fine-tuning with ONNX/TensorRT acceleration for real-time inference.

PythonOpenCVPyTorchStable DiffusionRT-DETRTensorRT
GitHub
2024-2025Featured

T.C. Ministry of Industry and Technology: Parallel Programming in PostgreSQL

Benchmarked query performance and parallel speedup across two PostgreSQL servers using controlled synthetic datasets. Produced an article-format report with findings and recommendations for HAVELSAN.

PostgreSQLPythonLLAMAUbuntu
2024Featured

TEKNOFEST 2024 Health AI

Trained YOLOv8 to detect breast cancer lesions in mammography images with 95% accuracy. Used two Llama-based LLMs to correct radiology reports and fine-tuned a BI-RADS classifier that achieved 98% accuracy.

YOLOLLaMAComputer Vision
Certification
2024

PalmReaderAI

Palm Reader AI web application using Streamlit, leveraging Llama-3.2-90B Vision model for advanced palmistry analysis.

PythonStreamlitLLaMA 3.2
Live DemoGitHub
2024

Feature Engineering for ML

Optimizing datasets using variance thresholding, scaling, encoding categorical variables, and dimensionality reduction.

PythonScikit-learnPandas
GitHub
2024

Spark California Housing Regression

Apache Spark MLlib pipeline for distributed regression on California Housing dataset. Achieved R² 0.797 and RMSE 44,570.

Apache SparkMLlibPython
GitHub
2023-2025

AI-Assisted Laboratory Animal Monitoring

Processed multi-camera video feeds with OpenCV and trained YOLOv8 for high-accuracy rat detection. Built a PyQt5 desktop interface for real-time experiment monitoring.

YOLOv8OpenCVPyQt5
2024

YOLO11N Traffic Sign Vision

Traffic sign detection using YOLO11N architecture with training logs, model weights, and sample predictions.

YOLO11Computer VisionPython
GitHub
2023

Ear Analysis Hybrid Model

Hybrid ResNet-CNN model for gender, identity, and age classification using ear photographs. Achieved 91% accuracy.

ResNetCNNTensorFlow
See more on GitHub

Capabilities

AI & LLM Systems

Multi-Agent Systems, RAG, Tool Calling, LLM Architecture, Fine-Tuning, LangChain, LangGraph

ML & Generative AI

PyTorch, TensorFlow, YOLO, RT-DETR, Computer Vision, FLUX.1-dev, TTS, Voice Cloning

Full-Stack & Data

Python, FastAPI, REST APIs, React, SwiftUI, JavaScript, Supabase, PostgreSQL, pandas

Infrastructure & Automation

Docker, Linux, MCP, Colab/GPU, Self-Hosted LLMs, n8n, Beszel, Power BI

Applied ML & Analytics

scikit-learn, NumPy, feature engineering, ETL, Gradient Boosting, model evaluation

AI Developer Tooling

Custom MCP servers, reusable coding skills, remote GPU workloads, experiment automation

PythonFastAPIReactSwiftUISupabasePostgreSQLLangGraphAgnoRAGTool CallingPyTorchTensorFlowYOLORT-DETRFLUX.1-devTTSpandasScikit-learnDockerLinuxMCPn8nPower BIPythonFastAPIReactSwiftUISupabasePostgreSQLLangGraphAgnoRAGTool CallingPyTorchTensorFlowYOLORT-DETRFLUX.1-devTTSpandasScikit-learnDockerLinuxMCPn8nPower BI

Certifications

TEKNOFEST 2024 Finalist

Health AI Competition, Disease Detection via Computer Vision

Agentic AI and AI Agents: A Primer for Leaders

Vanderbilt University (Mar 2025)

Milli Teknoloji Akademisi Yapay Zeka Uzman Eğitim Sertifikası

T.C. Sanayi ve Teknoloji Bakanlığı (June 2025)

Introduction to Docker

Google (Jan 2025)

IBM Deep Learning & Neural Networks with Keras

IBM (Sep 2024)

Academic & Volunteer

Development of a Modular Interface for AI-Supported Animal Behavior Experiment Software

ISBN: 978-625-6879-38-6

Google Developer Groups Core Team Member

GDG Kütahya - Community events, workshops, and AI-focused initiatives

Writer on Medium

@barisgulecdev

Speaker at Kütahya Innovative Medical Technologies and AI Congress 2024

Presenting on LLMs, RAG, and Transformer architecture

Selected for Harvard-affiliated Aspire Leaders Program

Let's build
something great.

Open to applied AI product collaborations and engineering opportunities. Let's discuss what we can build together.

Email Me
LinkedInGitHubMedium

baris@gulecbaris.tech