26.10.2026 aktualisiert

**** ******** ****
Premiumkunde
100 % verfügbar

LLM Engineer / Data Scientist

Wien, Österreich
Weltweit
PhD in Computer Science (Dr.)
Wien, Österreich
Weltweit
PhD in Computer Science (Dr.)

Profilanlagen

dr-wasif-masood-llm-engineer.pdf

Über mich

Dr Wasif Masood, founder of Empirisch Tech, is an AI engineer with a PhD from Klagenfurt and 19+ years in computer science, 10+ in ML. He specializes in LLMs, RAG, multi-agent systems and computer vision. He built T-Mobile Austria's first data science team, cutting churn by 17%.

Skills

Künstliche IntelligenzPythonMachine LearningNatural Language ProcessingAzure Machine LearningTransformerData ScienceLarge Language ModelsPrompt EngineeringDeep LearningMachine Learning OperationsArtificial Intelligence Markup Language (AIML)Daten-Pipeline
LLM Pretraining & Fine-Tuning
Umfangreiche Erfahrung im Pretraining und Fine-Tuning von Large Language Models (LLMs), einschliesslich Methoden wie LoRA/PEFT, QLoRA, DPO, ORPO sowie Retrieval-Augmented Generation (RAG) und Multi-Agent-Systemen.

Machine Learning & Deep Learning
Tiefgreifende Kenntnisse in Machine Learning und Deep Learning, inklusive Transformer-Architekturen, Bayesian Optimization, Convolutional Neural Networks, Object Detection und wissenschaftlichem ML / Surrogate Modeling mit Frameworks wie PyTorch und TensorFlow.

Cloud & MLOps (AWS, Azure, GCP)
Fundierte Kenntnisse in Cloud-Plattformen wie AWS, Azure und GCP, einschliesslich MLOps-Tooling mit CI/CD, Model Pipelining, Deployment und Scheduling sowie Diensten wie AWS Lambda, Redshift, Azure Data Factory und GCP Machine Learning.

Computer Vision & Synthetic Data
Erfahrung im Aufbau von End-to-End-Synthetic-Data-Pipelines in CARLA und Unreal Engine 5.5, Training und Benchmarking von Objektdetektoren (YOLOX, CSPDarknet, MobileNetV3) sowie Sim-to-Real-Transfer mit starken Ergebnissen.

Python & SQL Programmierung
Starke Programmierkenntnisse in Python und SQL, einschliesslich Frameworks wie Django, FastAPI, Ray sowie Datenbankloesungen wie PostgreSQL und ChromaDB.

NLP & Prompt Engineering
Kenntnisse in Natural Language Processing, Prompt Engineering, RAG-Architekturen, LangChain, Hugging Face Transformers sowie Evaluierungsmetriken wie ROUGE, SuperGLUE und HELM.

Edge AI & Model Compression
Kenntnisse in Edge-AI-Deployment, ONNX-Konvertierung, DeepSpeed, PEFT sowie Modellkompression und Datenaugmentation fuer effiziente Inferenz auf ressourcenbeschraenkten Systemen.

Data Engineering & Feature Stores
Erfahrung in der Entwicklung von Feature Stores, Automatisierung von ML-Pipelines, ETL-Prozessen sowie dem Einsatz von AWS MLOps-Stacks fuer vollstaendige Modell-Deployments.

Continual Learning & Foundation Model Adaptation
Kenntnisse in Continual Learning, Experience-Replay-Pipelines zur Vermeidung von katastrophischem Vergessen sowie Foundation-Model-Fine-Tuning und Neural Operators.

Erfahrung

Berufserfahrung5 - 10 Jahre
Freelancer seit2019

Sprachen

DeutschgutEnglischMuttersprache

Projekthistorie

LLM Scientist - Foundation Models & Continual Learning (Freelance)

LUMEIK5G
Adapted a 157.7M-parameter transformer-based foundation model (shifted-window / Swin neural operator) to new domains in PyTorch, implementing full fine-tuning, LoRA (PEFT), frozen-trunk, and multi-task joint training. Built a low-weight experience-replay training pipeline to mitigate catastrophic forgetting. Applied orthogonal decomposition (POD) to estimate dataset redundancy and intrinsic dimensionality. The model achieved 8.3x lower error than an equally budgeted model.

ML Scientist - Computer Vision (Freelance)

LUMEIK5G
Built an end-to-end synthetic-data pipeline in CARLA / Unreal Engine 5.5 with domain randomization and automatic bounding-box annotation. Trained and benchmarked 180+ compact object detectors (YOLOX-Tiny family) in PyTorch. Achieved strong sim-to-real transfer with 96.8% mAP@50 using synthetic-only training.

GenAI Consultant (Freelance)

Neolytica - A QPharma Company, USA
Fine-tuned LLMs using DPO and ORPO to enhance document-tagging accuracy in the medical domain. Implemented a RAG-based chat model for natural-language database queries. Customized LLMs for medical use cases including data extraction from PubMed and continued pretraining on a domain-specific corpus. Achieved 5% higher accuracy than GPT-4o in domain-specific topic tagging and sentiment analysis.

R&D Consultant on S2ST, TTS, ASR (Freelance)

Qonda GmbH, Germany
Wrote two research proposals for Investitionsbank Berlin (IBB): flow normalization using multi-stage single-channel speech enhancement approach in ASR, and optimizing S2ST with model compression and data augmentation for efficient speech-to-unit translation.

GenAI Consultant (Freelance)

NDI GmbH, Switzerland
Implemented a RAG-based chat model to help resolve customer queries for an online gambling platform.

GenAI Consultant (Freelance)

Confidential Client
Led a team of 3 to build a specialized chatbot for an insurance company. The bot functions as an expert, efficiently addressing commonly asked questions from internal employees, saving valuable time for experts who would otherwise need to respond to repetitive queries.

LLM Researcher (Freelance)

Confidential Client
Wrote a review of LLM concepts including LLM architectures (Autoencoder/Autoregressive/Seq-2-Seq), domain adaptation, Chinchilla scaling laws, exploring LLMs on the Hugging Face Hub, prompt engineering, fine-tuning with PEFT (QLoRA, prompt tuning), evaluation metrics (ROUGE, etc.), and benchmarks (SuperGLUE, HELM, etc.), RLHF.

ML Scientist (Freelance)

Max Planck Inst. for Plasma Physics
Implemented an ML framework to facilitate the process of scientific discovery. Implemented several explainable AI and causality concepts. Focused particularly on the scalability of ML algorithms.

Full Stack Data Scientist (Freelance)

IU Uni. Munich, Germany
Developed a feature store to speed up the model development process. Automated machine learning models using the complete AWS MLOps stack, including CI/CD, model pipelining, deployment, and scheduling. Built a model for text labeling to better understand student feedback.

Data Science Manager (Fixed)

T-Mobile Vienna, Austria
Set up the first data science department at T-Mobile Austria. Led a team of 6 data scientists, driving AI adoption across marketing, finance, and network operations. Projects included Cross-Selling (ML models for customer targeting, product matching, and channel selection; 17% campaign success rate increase), Up-Selling (ML models for ARPU increase; 28% campaign success rate increase), Lift and Shift (migration to GCP including ETL), Household Identification (model to identify customers in the same household), and Proactive Maintenance (time-series analysis and ML/deduction-based models to detect network problems; ticket resolution time reduced by half).

Data Scientist (Fixed)

T-Mobile Vienna, Austria
Implemented, deployed, and tested data science use cases. Projects included Tariff Optimization (portfolio gap analysis and shadow product analysis for retention), Value-based Network Rollout (ML model for network investment targeting; savings of several hundred thousand euros), Customer Experience (models for NPS, bill shock, network quality, service line interactions), and Churn Prediction (churn model with SHAP analysis; 17% reduction in churn after A/B testing).

Embedded Application Engineer (Fixed)

Samsung SDI Graz, Austria

Researcher (Fixed)

University of Klagenfurt, Austria

Software Engineer (Part-time)

Ericsson Eurolab, Kohlscheid, Germany

Software Engineer (Fixed)

TRG Software House, Lahore, Pakistan

Software Engineer (Fixed)

WATEEN Telecom, Lahore, Pakistan

Kontaktanfrage

Einloggen & anfragen.

Das Kontaktformular ist nur für eingeloggte Nutzer verfügbar.

RegistrierenAnmelden