Track record

Under the hood

Hardcore engineering.

Big systems. Hard limits. Work that holds up.

Under the hood

Selected
results.

These results come from prior engineering, advisory, and contracting roles.

RubyConf 2019 / Speaker

How do you make distributed systems fast and reliable under flaky network conditions?

My RubyConf talk on statistically optimal API timeouts.

Watch the talk

01 / Query optimization

3 days → 2 minper report

30+ collections. Thousands of field combinations.

A query optimizer that did only the work each report needed.

Define · Build · Check · Learn
  1. 01

    Define

    Reports crossed 30+ data collections and thousands of field combinations. Repeated database work and memory failures made them take up to three days.

  2. 02

    Build

    I built a query optimizer that chose only the steps each report needed.

  3. 03

    Check

    Reports finished in about two minutes. The result gave teams answers without a days-long wait.

  4. 04

    Learn

    Make the work fit the question. A flexible report should not have to run every possible query.

02 / Infrastructure

~$15k / moinfrastructure costs removed

Predictive scaling. Zero-write tracking.

Two changes across different layers of the system cut the monthly bill.

Define · Build · Check · Learn
  1. 01

    Define

    Reduce the cost of running the platform by changing how it used servers and stored data.

  2. 02

    Build

    I built predictive scaling to match servers to demand, plus a link-tracking system that skipped database writes.

  3. 03

    Check

    Together, the changes cut the infrastructure bill by about $15,000 a month.

  4. 04

    Learn

    Look across the system. Removing work at one layer and matching capacity at another can add up.

03 / Feed performance

300 → 5database queries per request

Rebuilt the backend. Cut the waiting.

Feed p95 fell from 2s to 200ms through query design, indexing, and caching.

Define · Build · Check · Learn
  1. 01

    Define

    A single feed request triggered about 300 database queries. The product feed’s p95 response time reached two seconds.

  2. 02

    Build

    I rebuilt the backend, improved indexing, and added caching. The query count fell to five.

  3. 03

    Check

    The feed’s p95 response time fell to 200 milliseconds. Shoppers could browse a more responsive feed.

  4. 04

    Learn

    Fix how the work flows through the system. More servers are not the only way to make it faster.

What I’m great
at building.

AI systems, business software, and the infrastructure behind them.

Applied AI & classifiers

I build and train in-house classifiers: AI models that sort documents, route work, or flag items for review.

Keep your data private. Leverage small judgments you can trust.

Distributed Systems Expert

I build high-throughput distributed systems. My work includes a platform handling tens of millions of real-time events a day through a workflow engine. I make it boring.

Capacity planning, tenant limits, and high-throughput pipelines.

Payments & commerce at scale

I build and improve systems where a failed request can mean a lost sale. I led development on a payments team moving over $1B a month. My timeout work helped prevent millions of failed payments.

Data systems & business intelligence

I build the data systems a business runs on—from complex reporting engines to the pipelines leaders use to understand performance.

Experimentation & observability

I make complex product behavior visible. I uncovered conflicting experiments, then shipped a shared framework for experimentation and observability.

Real-time software & advanced media

I build custom real-time conferencing software and sophisticated media-processing systems.

Have a different kind of system in mind?

Talk through what you need

Recommended / A 30-minute call

What could your team
build

Let’s discover the bottleneck, talk through a plan, and see where I can help.

Talk with Daniel

30 minutes, directly with Daniel. Bring your goal and what you’ve tried.

Have something special?
Email me instead.

Dan@0to1.dev

I typically respond to inquiries within 4 hours.

Add a few details (optional)

Daniel Ackerman

You work directly with me.

Your next move / Let’s talk

Let’s map your
next move.

30 minutes with me. We’ll talk through your goal, where AI fits, and a first step.

Start with a conversation.

You don’t need to choose a service first. We’ll work out where to start.

Choose a time

30 minutes · Bring your goal and what you’ve tried

Anything you want me to know? Optional

A short, high-level note is enough. Your note goes to Cal.com when you choose a time, or into your email draft.

Have something special in mind?

Email me instead

I typically respond to inquiries within 4 hours.

Read the free guide