Tech Stack
Tools, frameworks, and data engines used across production deployments and applied research projects — categorized by architectural role, with references to active implementations.
Production AI, Agents & Applied MLOps
The layer where determinism and observability matter most: agent graphs are supervised, traced, and evaluated before anything reaches production.
| Role | Tools | Where I use it |
|---|---|---|
| Agentic frameworks & runtimes | LangGraph, Pydantic AI, LangChain, LlamaIndex, FastAPI | Stateful multi-agent supervision in Agent Fabric |
| Model access & orchestration | Model Context Protocol (MCP), Vertex AI (Model Garden, Pipelines), Gemini, Anthropic Claude, Hugging Face, LiteLLM | Cost-aware inference routing in Inference Fabric |
| Inference serving & safety | SGLang, vLLM, Ollama, Llama Guard 4 | Local-first serving in Inference Fabric; safety filtering in Agent Fabric |
| Vector & graph storage | Qdrant, pgvector, LanceDB, KuzuDB, Pinecone, Chroma | Semantic caching and hybrid Graph RAG retrieval in Inference & Data Fabric |
| Evaluation & benchmarking | LangSmith, Arize Phoenix, DeepEval, Braintrust | Regression testing and multi-judge pipelines in Eval Fabric |
| LLM observability & FinOps | OpenTelemetry (GenAI semantic conventions), Langfuse, Helicone | Token, latency and cost telemetry in Inference Fabric |
Data Engineering & Lakehouses
Local-first by default: prototyping on embedded engines, then exporting to cloud lakehouses with no code changes — the working model validated in Emerging Tech.
| Role | Tools | Where I use it |
|---|---|---|
| Processing engines | Polars, DuckDB, Apache Arrow, Apache DataFusion, Apache Spark (PySpark), Databricks, Pandas | Vectorized execution in Data Fabric; local-first verdict in Emerging Tech |
| Streaming & CDC | Apache Kafka / Redpanda, Debezium, Google Pub/Sub, Dataflow | Streaming ingestion in Data Fabric and the client engagements in About |
| Orchestration | Dagster, Apache Airflow (Cloud Composer / MWAA), Prefect | Batch/streaming pipelines described in Collaboration |
| Storage & table formats | Apache Iceberg, Delta Lake, Apache Parquet, Cloud Storage, Amazon S3 | Open-table-format integration in Data Fabric |
| Transformation & modeling | dbt-core, SQL, Kimball dimensional modeling, Data Vault 2.0 | Automated dbt remediation in Engineering Fabric |
Cloud & Platform Infrastructure
Everything reproducible from code, policy-checked before it applies, and portable across the two clouds I work in.
| Role | Tools | Where I use it |
|---|---|---|
| Cloud providers | Google Cloud Platform (GCP), Amazon Web Services (AWS) | Reference architectures in Production Blueprints |
| Infrastructure as Code | Terraform (modular landing zones), Terragrunt, OPA Rego, Conftest | Multi-cloud foundations in Production Blueprints; policy-as-code spike in Emerging Tech |
| Containers & orchestration | Docker, Kubernetes (GKE / EKS), Google Cloud Run, Amazon ECS / Fargate | Containerized runtimes across Active Labs |
| GitOps & secrets management | ArgoCD, HashiCorp Vault | GitOps reconciliation and zero-trust credentials in Platform Fabric |
| Platform observability | Prometheus, Grafana, Mimir, Loki, Tempo | Centralized SRE telemetry in Fabric Ops |
| CI/CD & local toolchain | GitHub Actions, GitLab CI, just, uv | The toolchain this site runs on |
Languages, Databases & Standards
| Role | Tools | Where I use it |
|---|---|---|
| Programming languages | Python, SQL, Rust, Scala, Bash | Rust for the high-performance router in Inference Fabric; Python everywhere else |
| Databases | PostgreSQL, DuckDB, MySQL, Redis, BigQuery, Snowflake, Redshift | Portable export paths validated in Emerging Tech |
| Engineering standards | Data contracts, schema evolution, idempotency, pre-commit quality gates, Conventional Commits, SemVer | Applied across every project, including this repository |
Tools under active evaluation — with objectives, findings, and adopt/hold verdicts — live in Emerging Tech.