Skip to content
Taha smiling in front of the Brandenburg Gate in Berlin
Berlin

Hi, I'm Taha.

I study what neural networks represent, and build the systems around them. Visiting Researcher at DFKI, working on interpretability and representation learning.

Selected projects

Software I built on my own. Papers are under Publications, company work under Experience.

All projects

Experience

Research and engineering in parallel, from freelancing to DFKI.

Full experience
  1. Jun 2026 to nowNow

    AI and Software Consultant · Multiple contracts

    AI agents, self-hosted models and edge ML for production systems.

  2. Jun 2024 to nowNow

    Visiting Researcher · German Research Center for Artificial Intelligence (DFKI)

    Representation learning, interpretability and semi-supervised remote sensing.

  3. Nov 2025 to May 2026

    Research Associate · National Center for Artificial Intelligence (NCAI)

    Concept-based interpretability (E-TCAV), plus a personal project on adversarial neural cellular automata.

  4. Jan 2025 to May 2025

    Backend Engineer · Vyro.ai

    The ImagineArt developer platform and low-latency generative AI streaming.

  5. Apr 2022 to Jan 2025

    Freelancer · Fiverr and direct clients

    50+ full-stack projects for clients, Level 2 seller.

  6. Mar 2024 to Sep 2024

    Open-Source Contributor, Google Summer of Code · CVAT

    A configurable keymap system for an annotation platform; 8 PRs merged.

  7. Jan 2024 to Jun 2024

    Software Engineer · Thinksoft

    Co-developed WebbsAI, a platform that turns prompts into complete websites.

  8. Jun 2023 to Jan 2024

    Software Engineer · OneScreen

    An LLM presentation maker, plus agent and RAG prototypes.

Education

  1. Nov 2021 to Jun 2025

    BS Computer Science · National University of Sciences and Technology (NUST)

    Thesis at DFKI on semi-supervised slum segmentation from satellite imagery.

Publications

Research & papers
  1. arXiv:2602.04525 · 2026Preprint

    SLUM-i: Semi-Supervised Learning for Urban Mapping of Informal Settlements

    M. T. Mukhtar, S. M. A. Kazmi, K. Naseem, M. A. Chattha, A. Dengel, S. Ahmed, M. N. Bajwa, M. I. Malik

    A confidence-scaled pseudo-label threshold for semi-supervised segmentation of informal settlements, with a new Lahore dataset and spatial splits across three cities and two sensors.

    Project pagearXivPDFCode

  2. arXiv:2605.10261 · 2026Preprint

    E-TCAV: Formalizing Penultimate Proxies for Efficient Concept Based Interpretability

    H. Aslam, M. A. Chattha, M. T. Mukhtar, M. I. Malik, A. Dengel, S. Ahmed

    Fast proxies for concept-based interpretability: approximating TCAV scores from penultimate-layer representations across architectures and datasets.

    arXivPDF

Events

Selections and awards

  1. 2024

    Google Summer of Code

    Selected for CVAT, 1 of 1,220 contributors from 43,984 applicants.

  2. 2024 and 2025

    DAAD research funding

    Funded my research internships at DFKI. In 2024, 1 of 2 selected from about 250 students.

  3. 2024

    SOFTEC, 2nd place

    Second of 64 teams nationwide in a 24-hour software hackathon.

  4. 2025

    High Achiever Medals

    Three medals from NUST for national and international co-curricular achievements.

Let's work on something.

Research collaborations, PhD positions, or hard engineering problems.

tahamukhtar20@gmail.com