Maksim Kleshchenok

Principal/Staff ML & MLOps Architect. I build high-performance AI platforms and real-time pipelines.
Full CV (RU/EN): HH • LN • Email

Focus: LLM/ML platforms (ClearML, DataHub, MinIO), GPU perf (FP8/INT4, TensorRT/Triton, NVLink/IB, k8s), realtime voice/video (WebRTC/LiveKit, low-lat TTS/ASR).
Value: measurable latency/cost cuts, reproducible pipelines, incident-ready ops culture.

Selected hacks work (KvadricepsAI)

• Low-Poly 3D Generator — text/image → GLB/UV/PBR in ~1 min. End-to-end demo (GameDevAI 2025 — 1st). [repo]
• FetalCare CTG — ingest → analytics → clinician UI; rules+ML (NICE/ACOG/NICHD) (LCT 2025). [repo]
• robust face-ID — adversarial/deepfake-resistant pipeline, eval & docs Kryptonite ML’25 — 5rd. [repo]

Stack (short)

Python • TensorRT/Triton • Kubernetes • Docker (distroless) • ClearML • ClickHouse, Cassandra, Postgres • MinIO/S3 • Spark/Polars • FastAPI/Granian • Go • Rust

Metrics