CV

Contact Information

Name A. Martin Gökçü
Professional Title CS & Engineering Student
Email martingokcu@gmail.com

Professional Summary

Computer Science master’s student at Politecnico di Milano with experience building AI and workflow automation tools in business environments. I work mainly with Python, LLM systems, Docker, and PostgreSQL. My recent work includes agentic data systems, retrieval-based applications, and automation pipelines for finance workflows.

Experience

  • 2025 - 2026

    Milan, Italy

    Process Automation Intern
    Koinos Capital Sgr
    • Built an automated financial metrics pipeline retrieving competitor sets, market data, and acquisition KPIs for portfolio evaluation of M&A targets, reducing analyst research preparation time by 3 hours per candidate.
    • Built an LLM agent that pre-screened M&A prospects against acquisition criteria, reducing manual evaluation time for the investment team.
  • 2023 - 2024

    Milan, Italy

    Junior Software Developer Intern
    Azimut Holding
    • Collaborated with a team of four engineers and financial advisors to design and develop a core banking application for Azimut Holding’s new banking division.
    • Built internal workflow automation libraries integrating Microsoft Graph and SugarCRM APIs, reducing manual data-handling overhead across the operations team.
  • 2023 - 2023

    Milan, Italy

    Junior Software Developer Intern
    Brachitek

Education

  • 2025 - present

    Milan, Italy

    Master of Science
    Politecnico di Milano
    Computer Science and Engineering — T2A Track
  • 2022 - 2025

    Milan, Italy

    Bachelor of Science
    Politecnico di Milano
    Computer Science and Engineering (prev. Mechanical Engineering 2020–2022)

Projects

  • iData — Document Intelligence Platform

    Python, Flask, React, Milvus, MongoDB, PostgreSQL, Docker, LangChain

    • Built a self-hosted enterprise document intelligence platform that automatically ingests, parses, chunks, and indexes files dropped into a watched folder or synced from cloud storage (Google Drive, S3).
    • Multi-agent RAG backend (Orchestrator → Document Agent + SQL Agent) answers natural-language questions with inline citations, SQL lookups, and interactive chart generation through a React frontend.
  • ambercow.com

    Flask, Docker, LangChain, ChromaDB

    • Built and deployed a news platform that scrapes and clusters content from 50+ international outlets and generates summaries and QA through a custom RAG pipeline.
  • FakeLLaVA — Lightweight Vision-Language Model

    Python, PyTorch, Hugging Face Transformers, CLIP

    • Built a LLaVA-style VLM by connecting a CLIP ViT-B/32 vision encoder to Qwen2.5-0.5B through a two-layer projection MLP, trained in two stages (projection alignment, then full fine-tune) on CC3M and LLaVA-Instruct-150K.
    • Trained on Colab A100 (~6 h); published a blog post documenting the architecture and training process.
  • COT-RAG Retriever Agent

    Python, Unsloth, LangChain

    • Fine-tuned a language model using Unsloth to act as a chain-of-thought retriever inside a RAG pipeline, enabling structured reasoning over retrieved passages before generating answers.
  • LLM Fine-Tuning for TTS (D-Mel)

    Python, PyTorch

    • Developed a data generation pipeline downloading YouTube videos, performing speaker diarization, and converting audio into D-Mel tokens for TTS fine-tuning. Evaluated LoRA-based and full fine-tuning approaches.
  • Voice Extraction & Dataset Builder

    Python, Whisper, pyannote.audio

    • Pipeline that diarizes audio, transcribes each speaker segment with Whisper, and packages the output as a labeled dataset ready for TTS or ASR fine-tuning.
  • ANN-OY Vector Search Engine

    C, SQLite3

    • Built an on-disk vector search engine in C supporting fast cosine-similarity search over large embedding datasets.
    • Integrated SQLite3 for ACID-compliant metadata storage and implemented binary serialization/deserialization for persistent index reuse.
  • RISC-V Simulator

    Python, Streamlit

    • Implemented a RISC-V instruction set simulator in Python covering arithmetic, logical, shift, memory, and branch instructions, with an interactive Streamlit UI for step-through execution.
  • Contact Card Sharing App

    Flutter, Flask, Jinja2

    • Full-stack mobile app built with Flutter (GetX) and a Flask backend; supports user search, contact card customization, follow system, and REST-polled messaging.
  • Brainfuck Interpreter

    C

    • Built a Brainfuck interpreter in C — tape-based execution with bracket-matching functions for loop jumps.

Skills

Languages (): Python, C, C++, TypeScript, Java
Frameworks (): Flask, LangChain, Apache Kafka
Developer Tools (): Docker, Git, CI/CD
Databases (): PostgreSQL, MySQL, MongoDB
Libraries (): PyTorch, NumPy, Pandas, Matplotlib

Languages

Italian : Native
Turkish : Native
English : Advanced (IELTS: 8.5)