u
ulrich_cervatti

Ulrich C

@ulrich_cervatti

Ulrich DataLab

Brazil
English, Portuguese
About me
Systems & Data Engineer operating an on-demand multi-GPU cloud pipeline (RTX 4090 / A100 + 64-Core EPYC nodes). I specialize in turning massive, slow, or messy files into clean, structured deliverables with ultra-fast turnaround: - Batch PDF OCR & Table Extraction (Docling / Vision-LLM -> Excel, CSV, Markdown, JSON) - Bulk Audio/Video Transcription & Speaker Diarization (WhisperX -> SRT, DOCX, TXT) - Multi-GPU 3D Blender Cycles Rendering & Drone Photogrammetry (OpenDroneMap) - Large-Scale Web Scraping, DuckDB/Parquet Pipelines & RAG Vector Embeddings... Read more

Skills

u
ulrich_cervatti
Ulrich C
Offline • 

See my services

Automations
I will extract tables and text from scanned pdfs to excel via GPU
From$20 / projectMore details
Graphics & Design
I will process drone photos or video to 3d gaussian splatting ply or orthomosaic
From$45 / projectMore details

Work experience

Self_employed / Freelance

Data & Cloud Compute Engineer

Self employed / Freelance • Self-employed

Jan 2022 - Present • 4 yrs 9 mos

Architected and operated high-throughput automated data and GPU compute pipelines for B2B clients: • Processed 500,000+ complex scanned PDF pages, legal filings, and financial tables into structured Excel, CSV, Parquet, and Markdown datasets using GPU-accelerated OCR (Docling, Marker, Vision-LLMs). • Executed large-scale audio and video transcription and speaker diarization pipelines using WhisperX and Pyannote on NVIDIA RTX 4090 / A100 nodes. • Built resilient headless web scrapers and ETL pipelines using Python, Go, Playwright, and DuckDB capable of extracting and cleaning millions of records. • Provided multi-GPU headless 3D Cycles rendering (Blender OptiX) and drone photogrammetry processing (OpenDroneMap).