Low-level AI.
Real-world impact.
I design and build agentic AI systems and LLM pipelines for production. What sets my work apart: I understand the stack all the way down to CUDA kernels and GPU scheduling — so architectural decisions are grounded in what the hardware can actually do, not just what the framework makes easy.
▸ Agentic AI Engineer · LLM Architect · CUDA Specialist
▸ Agentic systems · LLM fine-tuning · Edge inference · MLOps/AWS
▸ Full stack: agent orchestration → low-level when needed
▸ true # open for new projects
// what I do
Services
End-to-end engagements from architecture design and PoC to production deployment and handover.
Agentic AI & LLM Orchestration
Multi-agent systems that plan, retrieve, and act autonomously. Tool-use pipelines, LLM-orchestrated workflows, memory and state management — built on solid infrastructure knowledge so security, latency and cost are designed in from the start, not patched in later.
LLM Fine-tuning & Alignment
Domain-specific model adaptation using LoRA, QLoRA, DPO, and RLHF pipelines. From data curation and preprocessing to production-ready model weights and evaluation frameworks.
Edge LLM Inference
Deploy quantized large language models on constrained or specialized hardware. Includes hardware selection for your use case and budget — through to custom CUDA kernels, GGUF/Open WebUI optimization, TensorRT integration, and kernel-level profiling.
MLOps & Agentic Architectures on AWS
End-to-end ML infrastructure and agentic architectures on AWS: SageMaker training & endpoints, Lambda-based inference, Step Functions for multi-agent orchestration, Bedrock integration, ECR, model monitoring, and GitOps CI/CD for ML teams.
HPC & CUDA Optimization
Squeeze maximum throughput from GPU clusters. Custom CUDA/ROCm kernels, tensor parallelism, flash attention variants, memory bandwidth optimization, NVLink & InfiniBand topology-aware scheduling, and systematic profiling with Nsight.
Audio & Signal Intelligence
Real-time audio ML pipelines: speech recognition, noise suppression, acoustic anomaly detection, and custom DSP preprocessing. Production deployments for embedded and cloud targets.
AI Strategy & Technical Leadership
ML background across LLMs, classical data science, computer vision, and signal processing — built over years of academic research and production work. This breadth allows me to assess the actual solution space: when a fine-tuned LLM is overkill, when XGBoost is the right call, and which specialists your team needs.
// client feedback
What clients say
"His work is exemplary. He is a prodigy. If he can't do it, 99.9999% of people can't. Don't think too much — he just delivers all the time."
— Client via Upwork
"The best deep learning engineer I worked with. After hiring multiple freelancers I have come to the conclusion that he is the best in the business."
— Client via Upwork
"The AI/ML project I'm working on is extremely novel and has been in progress for nearly 2.5 years, during which I've faced several long periods of no progress. Meeting and aligning with Alex has metaphorically been a one in a million encounter."
— Client via Upwork
"If he can't do it, nobody most likely can't. He is one of the best AI engineers that you can come across."
— Client via Upwork
"Excellent to work with — a prodigy that will go far."
— Client via Upwork
All reviews from verified client engagements on Upwork.
Top Rated Plus · Upwork profile// about me
The engineer behind the work
I design and build agentic AI systems and LLM pipelines for companies that need them to actually work in production — reliable, fast, and maintainable. Multi-agent orchestration, fine-tuned domain models, RAG architectures, AWS-native deployment: this is the core of what I do.
What makes the difference: I understand the full stack from cloud architectures down to GPU scheduling. That means I don't just wire together frameworks — I make the right architectural decisions because I know where the real bottlenecks are.
I work directly with engineering teams and technical founders, without layers of account management. You get senior expertise on the actual problem, fast iteration, and a clean handover.
Academic background
Distinction · Zemanek Prize nominee · 2018–2023
Distinction · 2015–2018
Distinction · 2011–2015
// tech stack
Languages
LLM & Agents
AWS
Cloud & Infra
Signal & Data
Frameworks & XAI
// let's work together
Have a project in mind?
I'm currently available for new engagements — consulting, full project builds, or embedded work with your team. Reach out and I'll get back to you within one business day.
Based in Vienna, Austria — available on-site (AT/DE) or remote worldwide
Preferred engagement: full project · consulting · embedded with your team