Now

What I'm doing now

Updated June 2025 · Inspired by Derek Sivers

Building

• → Data ingestion and quality validation frameworks for enterprise pipelines
• → Kubernetes-based job orchestration for multi-tenant workloads
• → Experimenting with local LLM inference pipelines using Ollama and llama.cpp
• → This website

Reading

• → Delta Lake paper — ACID on top of Parquet in object storage
• → Designing Data-Intensive Applications — Kleppmann (re-read)
• → DeepSeek-V3 technical report

Learning

• → Apache Flink internals — checkpointing, state backends, exactly-once semantics
• → vLLM and PagedAttention for efficient LLM serving
• → Rust — slowly, through Rustlings and small tools

Exploring

• → AI inference optimization: quantization, speculative decoding, KV cache strategies
• → eBPF for observability in Kubernetes workloads
• → OpenTelemetry adoption patterns in data pipelines