Machine learning, end to end.
I work across NLP, computer vision and applied ML, taking problems from dataset construction through model benchmarking to a documented, reproducible result. First author of a peer-reviewed computer vision paper (Sādhanā, Springer 2025). Python is my daily driver, with the OpenAI and Anthropic APIs for LLM tooling and agentic workflows.
Open to Werkstudent roles · from Aug 2026 · Saarbrücken / remote Now: reproducing MolCA (EMNLP 2023), graph-augmented LLMsReal-time yoga pose estimation with a hybrid MoveNet architecture
Built during a computer vision research internship at NIT Puducherry. The architecture splits deliberately: MoveNet Thunder generates high-fidelity keypoints for training, while MoveNet Lightning handles inference at 30+ FPS on consumer hardware: accuracy where the model learns, speed where the user waits. I owned the dataset end to end, consolidating and curating 3,435 images across 11 asanas from Kaggle and Yoga-82 sources, then trained the keypoint-embedding classifier, ran the misclassification analysis and carried the paper through two rounds of peer review.
MolCA reproduction: can an LLM read molecular graphs?
Fall 2026 graduate seminar on transformer language models and graph integration with Prof. Dietrich Klakow, Saarland University. My first-choice paper: MolCA (Liu et al., EMNLP 2023), which gives a text-trained LLM 2D graph perception by wiring a molecular graph encoder into Galactica through a Q-Former cross-modal projector, with a LoRA adapter for downstream fine-tuning. The coding task (roughly 60 hours) is a critical verification, not a demo: reproduce a downstream result, run the ablation behind the paper's central claim (does the graph modality actually help, or is the LLM doing the work alone?) and test generalization on data the authors didn't use. Culminates in a 10-page thesis-style report and a 35-minute final talk in September.
Lingomatch, AI document triage for a Gov/Justice-Tech platform
Student consultant (AI/Data Science) via the UdS Innovation Center's Triathlon startup programme: I built the ML backend of an AI-driven document-triage feasibility study for Lingomatch, turning photos of foreign-language documents into usable German gist translations for public authorities. I built and benchmarked the Arabic→German pipeline: Surya OCR evaluated on KITAB-Bench with CER/WER and confidence calibration; evaluated PaddleOCR as an independent second engine for disagreement-based flagging and documented its failure mode on Arabic due to engine-asymmetry; and benchmarked Opus-MT against NLLB-1.3B, matching translation quality (BLEU 43.8 vs 42.4) at a ninth of the memory, after catching the licence issue that ruled NLLB out for production.
Delivered the live demo in the final case-study presentation to the company's CEO.
DODI: deception scoring for Terms-of-Service documents
DODI (Digital Ownership Deception Index) quantifies the gap between "Buy now" marketing and "you're only licensing this" fine print in consumer ToS documents. A deterministic, fully reproducible scorer (Flesch–Kincaid readability, licence-vs-ownership term ratio, red-flag clause detection, weighted 25/50/25) validated against ToS;DR human expert grades (Spearman ρ = 0.54, n = 12) and applied as a temporal study: 10 platforms, four Wayback Machine snapshots each (2015–2024). The index passes its face-validity test: GOG, the DRM-free "you own your games" store, scores lowest of all ten platforms.
Shadow Correction: piracy as a market signal
Quantitative study (team of 3) testing whether piracy behaves as a market-correction signal, structured as three testable barriers. The primary parameter: streaming-market fragmentation, measured with the Herfindahl–Hirschman Index on Parrot Analytics demand-share data. The supporting parameters were designed as falsification tests: piracy-intent indices built from Google Trends (pytrends) aligned with World Bank PPP-adjusted hourly wages across 90+ countries, where the statistically-zero economic correlation (Spearman ρ = −0.001, p = 0.99) is itself the finding, isolating fragmentation rather than affordability as the driver; plus event-study analysis of piracy spikes after content delistings.
Build with
LLM work, daily
How I work
I benchmark before I recommend: models get tested on realistic inputs, and failure modes go in the report next to the wins; a translation model that degrades on long legal text is a finding, not a footnote. I take deployment constraints seriously from day one: model licences, GDPR, EU AI Act obligations and what actually fits in 8 GB of VRAM. Everything ships version-controlled and documented so it survives handover. Where I'm still building depth I say so and pick it up fast: Docker, CI/CD and cloud deployment were on that list until I shipped DODI's live scorer as a Dockerized FastAPI service with CI on every push.