Hi, I’m Furkan.
Founding Engineer / AI Product Engineer

Built and shipped an AI-first phishing and URL risk analysis SaaS/PaaS from zero to one — owning architecture, backend, cloud infra, browser extension, auth, security, and deployment. Designed a Google ADK-based agentic AI orchestrating 40+ Go microservices for multi-layered URL risk analysis and scoring.

Developed scalable infrastructure with Go, Google Cloud Run, BigQuery, Firestore, Kafka, and cost-optimized AI workflows. Implemented public/private APIs, real-time extension telemetry, auth flows, and integration-ready automation for customer-facing workflows.

Owned full productization including API-first access, usage-based pricing, CI/CD, demo environments, pilot deployments, and go-to-market execution.

Built an AI-native literature review SaaS from zero to one with Next.js, Go REST API, PostgreSQL, Redis, MinIO, and microservices. Designed a multi-LLM architecture supporting Gemini, Claude, OpenAI APIs, and local models via Ollama for search, RAG chat, extraction, and report generation.

Implemented end-to-end ingestion and RAG pipelines with LightRAG vector/graph hybrid retrieval, SSE streaming, semantic search, citation-grounded answers, and block-level evidence tracing. Built scholarly API integration, LLM-first extraction, and Google ADK multi-agent report generation.

Owned system design, backend, auth/security, and full product workflows from database to deployment.

Built a full-stack OCR SaaS from zero to one with Next.js, Go, PostgreSQL, MinIO, and Docker Compose. Designed async OCR workflows with file validation, job queue, polling, markdown rendering, credit-based billing, API key management, and webhooks.

Implemented a flexible provider architecture supporting LlamaParse, Mistral OCR API, and self-hosted models via Ollama, with Go worker pools and resilient background processing. Owned system design, CI/CD, and self-hosted production infrastructure.

Developed a production AI chatbot interface for natural-language security platform interaction — scan triggering, asset/vulnerability lookup via LangChain RAG with Qdrant, Supabase, FastAPI, and vector search. Improved alert accuracy by 8% through prompt engineering and model iteration.

Built and productionized ML detectors (Login Page, File Upload) and developed n8n automation workflows for pentest report validation, scanner comparison, and LLM-powered summarization.

Led applied AI/ML research across NLP, computer vision, Transformers, and cybersecurity. Built automated pipelines for large-scale data processing, model training, and evaluation. Created the MTLP-Dataset with 100K+ samples for ML research.

Authored peer-reviewed papers and a book chapter in applied AI, ML, computer vision, and cybersecurity. Translated research into applied prototypes and production-ready ML systems.

Assisted in delivering university courses on Data Intelligence and Statistical Analysis — data preprocessing, visualization, hypothesis testing, statistical analysis, and linear models.