JH.

Backend & data engineering

Systems that hold up when the work gets real.

I’m Jacob Haflett, a senior engineer building data platforms, distributed systems, and AI pipelines that stay useful under load.

See selected work Start a conversation
100K
requests/sec API scale
35K
robot fleet telemetry
99.9%+
real-time availability

Selected work

Technical work with a point of view.

Social platform · product systems

KITKAN

A local-first social platform for discovering and sharing work together. I am shaping the product, system contracts, and the infrastructure that supports it.

View project

Infrastructure · Kubernetes

Home Lab

A self-hosted Kubernetes environment designed to grow from lightweight services into a hybrid infrastructure playground.

View project

Go · interoperability

Universal Chat Bridge

An experiment in carrying conversations across messaging platforms, with voice and video support in view.

View project

Experience

Production systems, from data to delivery.

  1. 2025—now

    Full-Stack ML / AI Engineer · Verb (MARKET.live)

    Multi-tenant APIs and event-driven inference on TypeScript, Go, Python, AlloyDB, and ClickHouse—built for real-time personalization and analytics.

  2. 2023—2025

    MLOps / Cloud Engineer · Jabil / Badger Technologies

    Spark and lakehouse systems for 35,000 robots, across AWS, GCP, and Azure. Owned reliability, observability, and on-call delivery.

  3. 2022—2023

    DevOps / Cloud Engineer · AWS

    Operated data services and automated cloud infrastructure for customer workloads spanning batch, streaming, and Kubernetes delivery.

  4. 2021—2022

    Software Engineer · Air Liquide / Airgas

    Built Spark indexing and ETL pipelines across millions of SKUs, plus low-latency services for search and recommendation systems.

Capabilities

The stack is the means, not the headline.

Data platforms

Spark, Kafka, Pub/Sub, BigQuery, Snowflake, PostgreSQL, ClickHouse.

Distributed systems

REST, gRPC, multi-tenant services, queues, workers, retries, and recovery.

AI delivery

LLM-powered products, vLLM serving, evaluation, tracing, and event-driven inference.

Reliable operations

Kubernetes, Terraform, Docker, OpenTelemetry, Prometheus, Grafana, and practical on-call ownership.

Contact

Have a difficult system to make dependable?

jacobhaflett@icloud.com
GitHubNew York, USA