HM Hamid Matiny

Portfolio · AI platforms

Mohammadreza “Hamid” Matiny

AI Infrastructure & MLOps Engineer specializing in LLM serving, GPU orchestration, and production observability.

Currently: Torc Robotics · Data Quality Assurance ML Pipeline

GitHub LinkedIn Email

About

Platform engineering for models that have to ship.

I’m Mohammadreza (“Hamid”) Matiny — an AI infrastructure and MLOps engineer with ~6 years across deep learning systems, data pipelines, and cloud infrastructure on GCP and AWS. My work sits where model code meets the platform: serving contracts, GPU scheduling, experiment provenance, and the observability that makes production behavior trustworthy.

Today I build AI platform infrastructure — LLM serving, GPU orchestration, fleet telemetry — while also working on agentic-AI security (policy-as-code gateways, tool-abuse controls, audit trails). At Torc Robotics I contribute to the Data Quality Assurance ML Pipeline that keeps AV training data contract-clean before it ever reaches a model.

I care about honest maturity signals: tagged releases when they’re real, architecture diagrams that match the repo, and metrics that come from measured runs — not marketing copy.

~6 years Deep learning systems, data pipelines, and cloud infrastructure
LLM platforms Multi-backend serving, GPU cost-safety in CI, tool-grounded advisors
Production data Contract gates, lakehouses, streaming QA, and fleet observability

Featured work

Case studies, not just repos.

Architecture, engineering decisions, and honest maturity — verified against public GitHub sources.

Vulcan

Active on main

Multi-backend LLM serving and GPU-orchestration platform — one contract across vLLM, Triton, Ray Serve, KServe, and BentoML.

Tagged v1.0.0 at phase-15. Main has since completed phases 16–22 (advisor, training, LoRA/PEFT, DVC, tracking) — CHANGELOG records 1.2.0; no v1.1.0/v1.2.0 Git tag yet.

Architecture

Clients and a LangGraph advisor hit a routing gateway that speaks a unified model-serving contract (/health, /metrics, /v1/infer). Backends — BentoML, Ray Serve, Triton (+ TensorRT-LLM), vLLM (+ GPTQ/AWQ/FP8 packs), KServe — plug in behind that contract. Training jobs (Ray Train, FSDP/DDP, DeepSpeed, LoRA/PEFT) share a TrainingJobSpec; MLflow and W&B track runs; DVC versions exports. GPU infra (Kueue / Karpenter / MIG) is validated in CI without burning real GPU cost. SageMaker and Bedrock are selectable managed paths. Observability (Prometheus, Grafana, Tempo, cost-per-token) grounds every advisor number in real evidence.

Key decisions

  • Unified serving contract instead of per-backend APIs (ADR-001)
  • GPU cost-safety: validate-only infra in CI; no invented tokens/s (ADR-002, ADR-007)
  • Kueue multi-tenant scheduling + MIG partitioning strategy (ADR-003, ADR-004)
  • LangGraph advisor is tool-grounded — every stated number must appear in Prometheus/benchmark evidence (ADR-014)
  • Pluggable experiment tracking: MLflow self-hosted + W&B offline-only in CI (ADR-013)
  • vLLM
  • Triton
  • Ray Serve
  • KServe
  • BentoML
  • Kueue
  • Karpenter
  • TensorRT-LLM
  • LoRA/PEFT
  • DVC
  • MLflow
  • W&B
  • LangGraph
  • SageMaker
  • Bedrock

System shape

Vulcan architecture Clients and LangGraph advisor connect to a gateway and model-serving contract, which fans out to BentoML, Ray Serve, Triton, vLLM, and KServe. Training, GPU infra, and observability feed the same system. Clients console / API Gateway routing :9007 Advisor · LangGraph tool-grounded only Model serving contract /health · /metrics · /v1/infer BentoML Ray Serve Triton + TRT-LLM vLLM GPTQ/AWQ/FP8 KServe Training contract Ray · FSDP · LoRA Tracking + DVC MLflow · W&B GPU · Kueue/Karpenter/MIG Observability Managed SM · Bedrock

View repository →

Argus

Early build

Vulcan's sibling — unified fleet telemetry, data-quality, MLOps, and observability with an AI copilot.

Architecture / early-build phase. Pipeline topology and contracts are defined; not a finished production system yet.

Architecture

Kafka/Redpanda ingest → Ray processing → Flink QA gate → Iceberg lakehouse → Dagster orchestration → drift-monitor → OPA-backed incident engine → OpenTelemetry → dashboard, plus a read-only AI copilot. Same containers run via Docker Compose locally or Terraform/Helm/Argo CD on EKS.

Key decisions

  • Contract-first streaming path with an explicit Flink QA gate before lakehouse writes
  • Iceberg + Dagster for reproducible lakehouse materialization
  • OPA policy for incident decisions — not prompt-only automation
  • Copilot is read-only against telemetry and runbooks
  • One image set for local Compose and EKS GitOps
  • Kafka
  • Redpanda
  • Ray
  • Flink
  • Iceberg
  • Dagster
  • OpenTelemetry
  • OPA
  • Argo CD
  • Terraform

System shape

Argus pipeline architecture Streaming path from Kafka or Redpanda through Ray, Flink QA, Iceberg lakehouse, Dagster, drift monitor, incident engine, OpenTelemetry, dashboard, and AI copilot. Kafka / Redpanda Ray Flink QA gate Iceberg lakehouse Dagster Drift monitor Incident engine · OPA OpenTelemetry Dashboard AI copilot

View repository →

hydra-data-factory

Validated

Production-validated AV telemetry lakehouse — contract validation, DLQ isolation, Terraform-provisioned AWS path.

Production-validated pipeline with measured run results: 42,972 records ingested · 91.8% acceptance · 0% post-gate rejections.

Architecture

Ingest mock fleet JSON → Pydantic + Pandera contract validation with DLQ isolation for rejects → PyArrow/Parquet (Snappy, Hive partitioning) into S3 → Glue catalog. Dual orchestration: local Airflow + MLflow, and AWS Step Functions + Lambda. Infra fully provisioned with Terraform (S3, Glue, IAM).

Key decisions

  • Dual-path orchestration (Airflow local / Step Functions cloud) sharing one transformation core
  • Hard contract gate with DLQ isolation — rejects never contaminate the lake
  • Hive-partitioned Parquet for query-friendly AV telemetry
  • Everything provisioned as Terraform — no console-only resources
  • Pydantic
  • Pandera
  • PyArrow
  • Parquet
  • Terraform
  • S3
  • Glue
  • Airflow
  • Step Functions
  • MLflow

System shape

hydra-data-factory architecture Fleet JSON ingest through Pydantic and Pandera validation with DLQ isolation, then Parquet to S3 with Glue catalog, orchestrated by Airflow or Step Functions via Terraform. Fleet JSON ingest Pydantic + Pandera contract gate DLQ isolation Parquet · Hive S3 + Glue Airflow + MLflow Step Functions Terraform

View repository →

Also building

Security, edge vision, geospatial.

Active side systems — lighter treatment, same bar for accurate links.

aegis

LLM/agent security gateway — prompt-injection defense, policy-as-code, tamper-evident audit trails, human approval for high-risk actions.

AegisFlow

Multi-agent incident response on Temporal, PydanticAI, and OpenTelemetry — triage, plan with approval, verify, compensate.

edgevision

Edge CV anomaly detection for existing camera fleets — VMS-agnostic threat analytics (YOLO + ByteTrack lineage).

smart-assistant

Personal knowledge and workflow agent for local context and task automation.

terra-OBIA

Object-based image analysis for forestry stand delineation, wetland classification, and land-cover mapping at province scale.

terra-obia-etl

ETL for GeoNB / GNB open data — harmonized labeled stand polygons feeding terra-OBIA classifiers.

Foundations

Earlier systems that shaped the craft.

Grouped briefly — perception, serving, telemetry, and API foundations.

vit-fastapi-cloud-deploy ViT classification service on FastAPI + Docker + GCP object-detection End-to-end OD train → track → serve → CI/CD multi-modal-od Radar + LiDAR + camera fusion for AV detection cuda-optimization PyTorch optimization playbook (AMP, compile, MPS) lstm-attention-transformers Sequence models, attention viz, seq2seq workflows task-master FastAPI + PostgreSQL + Docker + Actions API bert-inference Containerized DistilBERT sentiment inference NEXUS-CV Real-time multi-modal CV platform with MLOps sentinel-ray Ray ingest + Pandera QA + drift + incidents vanguard-telemetry-monitor Fleet telemetry sim, anomalies, Prometheus/Grafana

Skills & stack

Where the work actually concentrates.

Organized by the systems I ship — not an exhaustive badge wall.

LLM Serving & Inference

  • vLLM
  • Triton Inference Server
  • TensorRT-LLM
  • Ray Serve
  • KServe
  • BentoML
  • GPTQ / AWQ / FP8
  • LoRA / PEFT
  • Bedrock
  • SageMaker

GPU Orchestration & Cloud

  • Kubernetes
  • Kueue
  • Karpenter
  • NVIDIA MIG
  • AWS
  • GCP
  • Terraform
  • Helm
  • Argo CD
  • Docker

MLOps & Experiment Tracking

  • MLflow
  • Weights & Biases
  • DVC
  • GitHub Actions
  • Dagster
  • Kubeflow
  • Prometheus
  • Grafana
  • OpenTelemetry

Data & Streaming Systems

  • Kafka / Redpanda
  • Apache Flink
  • Apache Iceberg
  • PyArrow / Parquet
  • Pandera
  • Pydantic
  • Airflow
  • Step Functions
  • Ray

Security & Agentic Systems

  • Prompt-injection defense
  • Policy-as-code (OPA)
  • LangGraph
  • PydanticAI
  • Temporal
  • Audit trails
  • Red-teaming patterns

Contact

Let’s talk platforms.

Open to conversations about AI infrastructure, LLM serving, MLOps platforms, and agentic security — especially roles where production reliability matters as much as model quality.

Available globally · open to remote

hamidmatiny@gmail.com LinkedIn