HEALTHCARE AI · COMPUTER VISION · EMBODIED INTELLIGENCE

I build AI that stays intelligent when the shortcut breaks.

I’m Zeeshan Modi—an AI/ML engineer and M.Sc. AIIS student in Bremen, working across robust medical imaging, robotics, neurotechnology and local-first AI systems.

ZM / SIGNAL-01LIVE
BASE
Bremen, DE
FOCUS
Robust AI
STATUS
Open to work
SCROLL TO EVIDENCE
69%
shortcut-gap reduction0.443 → 0.137
1st
AI in Health HackathonUBRA · Bremen
14
successful verification runsmulti-seed validation
OSS
merged MONAI + DFKI fixesproduction codebases

01 / SELECTED INTELLIGENCE

Not just models. Systems with evidence.

Selected from my GitHub and academic work: projects where the problem, engineering decisions and outcome can all be inspected.

FLAGSHIP 01HEALTHCARE AI / MODEL ROBUSTNESS

Pneumothorax Detection Without the Easy Shortcut

The baseline chest X-ray classifier could learn “drain = pneumothorax” instead of learning the condition itself. I treated that failure mode as the real problem—not just the leaderboard score.

BASELINE GAP0.443
ROBUST GAP0.137
−69%
INTERVENTION

2.5× minority-group WeightedRandomSampler with DenseNet121.

VALIDATION

14 successful runs across different seeds; 85.7% met the target gap.

RECOGNITION

1st place · UBRA / University of Bremen AI in Health Hackathon.

Inspect code, experiments and verification

SYSTEMS THAT SHIP

From capability to usable workflow.

02 / OPEN-SOURCE TRANSMISSION

Small patches. Real codebases.

Contributions across medical imaging, robotics simulation and computer-vision tooling—where changes need to fit systems other people already depend on.

Zeeshan ModiZM / BREMEN
N53°04′E8°48′2026

03 / PROFILE

Curiosity, engineered into useful systems.

I’m an AI/ML engineer who enjoys the full arc: finding the real failure mode, testing assumptions, building the system and making the result understandable to others.

My path has moved from computer science and AI in India, through an exchange semester in Ankara, to an M.Sc. in Artificial Intelligence and Intelligent Systems at the University of Bremen. Today I work across healthcare AI, computer vision, robotics, signal processing and local language systems.

I also teach Python and take on selected freelance work through Litends—now my independent AI and web practice, not an active startup.

2025 — PRESENTM.Sc. Artificial Intelligence & Intelligent Systems

University of Bremen · Germany

2021 — 2025B.E. Computer Science · AI & ML

Chandigarh University · First Division with Distinction

Download full résumé

Python Tutor · Student Assistant

HereAhead · University of Bremen

Led weekly programming sessions for M.Sc. students, designed practical exercises and documented course knowledge for continuity.

Machine Learning & AI Intern

Techlive Solutions

Fine-tuned deep-learning models and built BERT recommendation and NER pipelines exposed through REST APIs.

Independent AI & Web Freelancer

Litends · Freelance practice

Focused AI tools, automation and modern web applications. Litends began as a startup exploration and now serves as my independent practice.

04 / CAPABILITY MAP

The stack follows the problem.

I care less about collecting tool logos and more about combining the right methods into an inspectable, usable system.

P01

Perception

Medical imaging · OpenCV · DICOM · XAI · robustness testing · signal processing

L02

Language

Transformers · RAG · BM25 · dense retrieval · NER · local LLM systems

E03

Embodied systems

Robotics · PyBullet · ROS · planning · PID · particle filters · RANSAC

S04

Shipping

FastAPI · Docker · Streamlit · Gradio · MLflow · GitHub Actions · Linux

011st Place

AI in Health Hackathon · 2025

022nd Place

BRIDGE “Start it up! Gesundheit” · 2025

14Student Track

Bremen Big Data Challenge · 2026

Accepted papers

Cybersecurity + career recommendation · 2024

AVAILABLE / BREMEN / REMOTE / GERMANY

Have a hard problem worth making intelligent?

I’m open to Werkstudent, HiWi, internship and junior AI/ML roles, as well as selected freelance AI and web projects.