Skip to content

~/stack

Infrastructure stack & competency radar

Self-assessed across 10+ years of production execution. Hover any radar vertex for the evidence basis.

Competency Radar

8 axes · 0–100
Competency radar — 8 axes, scores 0–100Platform MLOps: 85 of 100. Dist. Systems (CAP): 82 of 100. Privacy & Compliance: 80 of 100. HPC Compute Opt.: 68 of 100. FinOps & Strategy: 78 of 100. DL Math & Internals: 65 of 100. Agentic Orchestration: 75 of 100. SWE & Data Eng.: 83 of 100 Platform MLOps 85/100 Dist.Systems (CAP) 82/100 Privacy &Compliance 80/100 HPC ComputeOpt. 68/100 FinOps &Strategy 78/100 DL Math &Internals 65/100 AgenticOrchestration 75/100 SWE & DataEng. 83/100

  • Platform MLOps · 85/100

    Production systems across 5+ orgs, upstream Kubeflow/TFx PRs

  • Dist. Systems (CAP) · 82/100

    K8s StatefulSets, Kafka, event-driven at Amber/Montu scale

  • Privacy & Compliance · 80/100

    0.87+ F1 PII redaction, RBAC, Privacy-by-Design in healthcare

  • HPC Compute Opt. · 68/100

    vLLM PagedAttention, FlashAttention, INT8/FP4 quantisation

  • FinOps & Strategy · 78/100

    40% cost reduction at Amber, Build vs Buy RFC authorship

  • DL Math & Internals · 65/100

    Custom autograd/attention, SVD/PCA derivations, Causal Inference

  • Agentic Orchestration · 75/100

    scaling-succotash production system, LangGraph/DSPy/MCP/A2A

  • SWE & Data Eng. · 83/100

    O(1) stdlib optimisation, upstream dask PR, full-stack TS/Python

Self-Assessment Methodology Note: Competency scores are self-assessed based on production execution breadth, not academic certification. Each axis includes a methodology string providing the evidence basis. Hover any radar vertex to see it.

Full Infrastructure Stack

7 categories

Languages & CS Primitives

  • Python (Asyncio, Metaprogramming)
  • C / C++
  • TypeScript
  • Advanced SQL (Window Functions, CTEs)
  • Bash / Shell
  • Perl

AI/ML & HPC

  • PyTorch
  • TensorFlow
  • Keras
  • Scikit-Learn
  • XGBoost / LightGBM
  • vLLM
  • FlashAttention
  • Quantisation (INT8 / FP4)
  • GPU Profiling (torch.profiler)

Agentic & GenAI

  • LangChain
  • LangGraph
  • DSPy
  • MCP
  • A2A (Agent-to-Agent)
  • RAG
  • GraphRAG
  • Guardrails
  • Prompt Engineering
  • Evals
  • Federated Learning
  • Differential Privacy

MLOps & Data Engineering

  • Kubeflow
  • ClearML
  • TFx
  • MLflow
  • DVC
  • Airflow / Prefect
  • Feast / Tecton
  • Kafka
  • BigQuery
  • Snowflake
  • PostgreSQL
  • Redis
  • Dask

Cloud & Infrastructure

  • GCP (GKE, Cloud Run, VertexAI, Pub/Sub)
  • AWS (EKS, SageMaker, Lambda, DynamoDB)
  • Docker
  • Kubernetes (StatefulSets, Operators, CRDs)
  • Terraform
  • Helm
  • CloudWatch
  • Datadog

Advanced DL Paradigms

  • PEFT (LoRA / QLoRA)
  • RLHF (PPO / DPO)
  • Transformer Architectures
  • Custom Autograd / Attention

Systems Architecture & Theory

  • Distributed Systems Theory (CAP, PACELC, Paxos / Raft)
  • Event-Driven Architecture
  • Domain-Driven Design (DDD)
  • Gang of Four Design Patterns
  • LSM / B-Trees