Author Image

Hello there, I am Michael

Michael Medford, Ph.D

Data Infrastructure & Production AI Consultant

Hands-on founding engineer and engineering leader with staff-level experience in data infrastructure and production AI. I help teams improve pipeline reliability, database performance, and cloud efficiency, and build dependable retrieval and AI orchestration systems. Available for remote consulting up to 10 hours per week.

Northwestern University: Bachelor of Science in Physics & Theatre
UC Berkeley: Ph.D in Astronomy & Astrophysics
DillyAI: Founding Engineer & VP of Engineering
BS: Physics & Theatre
Ph.D: Astronomy & Astrophysics
VP of Engineering

Technical Expertise

Experiences

1
Founding Engineer & VP of Engineering
DillyAI

June 2026 - Present, Remote

DillyAI is bringing AI to the back office of corporate accounting. We’re building the system that understands how accountants actually work, so firms can hand off the grunt work of an engagement to software that reasons about their numbers, not just reads them.

Responsibilities:
  • Built an event-sourced orchestration engine with Temporal for production LLM workflows, incorporating auditable verification pipelines
  • Built AI-assisted software development workflows with parallel execution, document concurrency controls, and automated cloud execution under human-directed architecture

Staff Software Engineer - Data Systems
KoBold Metals

April 2025 - May 2026, Berkeley, CA

KoBold Metals combines geology, machine learning, and large-scale software systems to reinvent how critical minerals are discovered for the global energy transition.

Responsibilities:
  • Built the Korecam360 data infrastructure for large-scale image ingestion, synchronization, and retrieval, supporting machine learning, historical backfills, and visualization.
  • Built an operational reporting application that reduced drill superintendent reporting from multiple hours each morning to approximately 15 minutes.
  • Led adoption of AI-assisted development practices, redesigned engineering workflows, and mentored engineers in AI-native development using Cursor and agent-based tools.
2

3
Staff Software Engineer - Subject Matter Expert in Datastores
Insight M

July 2024 - March 2025, Sunnyvale, CA

Insight M monitors oil and gas pipelines to detect methane leaks and prevent them from getting into the atmosphere, reducing the environmental impact of these operators.

Responsibilities:
  • Developed a data architecture and security roadmap covering development environments, analytics pipelines, and data retention, with projected annual savings of six figures
  • Optimized database processes by refactoring high-cost computational components, reducing runtime expenses, and enhancing system resiliency with comprehensive test coverage
  • Enhanced data infrastructure by implementing a database event alert pipeline, launching an internal application for query plan analysis to improve query efficiency, and introducing the company’s first postmortem template to foster learning from software retrospectives

Aumni, a JPMorgan Chase Company

November 2022 - June 2024, Salt Lake City, UT

Aumni gives you instant access to the most reliable financial and legal insights across your portfolio, sourced directly from your executed legal agreements.

Staff Software Engineer

July 2023 - June 2024

  • Redesigned backend architecture with management, product, and engineering to enable flow in stream-aligned teams
  • Developed a migration plan to maintain 100% uptime while refactoring a Ruby on Rails monolith into independent services within the JPMorgan ecosystem
  • Constructed contract testing framework to ensure independent deployability in collaboration with our Testing Enablement Team
Senior Software Engineer and Technical Lead

November 2022 - June 2023

  • Designed and deployed an OpenSearch indexing service for millions of documents in six weeks
  • Invented an SQL unit test framework supported by a file parsing library on top of our core business logic
  • Reduced developer data synchronization time from 15 minutes to 20 seconds through infrastructure and tooling improvements
4

5
Planet Labs

November 2020 - October 2022, San Francisco, CA

Planet provides daily satellite data that helps businesses, governments, researchers, and journalists understand the physical world and take action.

Technical Lead of Planet Fusion Monitoring Pod

March 2022 - October 2022

  • Increased daily Python processing capacity 10x while reducing per-unit costs 80% in four months
  • Reduced database query latency and CPU utilization 98% through SQL schema and query optimization
  • Reduced cloud data-delivery processing overhead 30x using batched GCP Pub/Sub messages
Geospatial Software Engineer

June 2021 - October 2022

  • Implemented automated data backups & disaster recovery via terraform to comply with ISO certification
  • Set up on-call notifications by integrating Sentry, PagerDuty and Slack into our Python & Go stack
  • Created a comprehensive cost estimate method to measure profit margin goals for our product teams
  • Coordinated several refactors across our Compute, Pipeline and Infrastructure teams to facilitate growth
Part-time Geospatial Software Engineer

November 2020 - May 2021

  • Enabled end-to-end development by connecting our React front-end, Go orchestrator & Python scripts
  • Conducted our largest scale test to date, giving the green light to onboard new customer contracts
  • Transitioned change detection algorithm to parallelized cloud context for a 100x speedup in E2E runtime

Founder & Tutor
Renaissance tutoring

September 2011 - June 2015, Chicago, IL

A tutoring agency that focused on communication, empathy, and creative thinking to help students gain fundamental understanding and successful problem solving strategies

Responsibilities:
  • Instructed students pursuing high school, undergraduate, and post-baccalaureate degrees in physics, chemistry, and mathematics (including AP, SAT, SAT II, ACT, and MCAT exams)
6

7
Co-founder, Managing Director & Actor
The Poor Theatre

September 2012 - June 2015, Chicago, IL

An ensemble based theatre company that presents high-quality and engaging productions from America’s up and coming new playwrights

Responsibilities:
  • Founded a company that worked with over 100 artists and had over 1000 audience attendees
  • Acted in over fifty stage productions and films throughout the Chicago-land area
  • Completed training at The Artistic Home & Improv Olympic (Chicago), and the British American Drama Academy (Oxford)

Education

2015-2021
Ph.D in Astronomy & Astrophysics
GPA: 3.79 out of 4
Extracurricular Activities:
  • Awarded $130,000 by winning the UC-National Lab In-Resident Graduate Fellowship at Lawrence Livermore National Lab
  • Lead instructor of graduate student mentorship program for 3 years
  • Lead instructor of required graduate student pedagogy training course for 3 years
BS in Theatre & Physics
GPA: 3.75 out of 4
Extracurricular Activities:

Projects

zort
zort
Owner Aug 2019 - Apr 2021

ZTF Object Reader Tool is a set of functions to organize and access the ZTF Public Data Release lightcurves across multiple colors.

fringez
fringez
Owner Nov 2020 - Apr 2021

Pipeline software used to remove atmospheric fringes from optimal astronomical images. Official stage in the Zwicky Transient Facility i-band data reduction pipeline running on 50,000+ images per night.

PopSyCLE
PopSyCLE
Developer Aug 2017 - Apr 2021

A Milky Way microlensing simulation code that stellar evolution, dust models and observation effects to identify optimal strategies for microlensing detections.

Publications

60 Microlensing Events from the Three Years of Zwicky Transient Facility Phase One

We conduct the first multiyear search for microlensing events on the Zwicky Transient Facility (ZTF), an all-sky optical synoptic survey that observes the entire visible northern sky every few nights. We discover 60 high-quality microlensing events in the 3 yr of ZTF-I using the bulk lightcurves in the ZTF Public Data Release 5. 19 of our events are found outside of the Galactic plane, nearly doubling the number of previously discovered events in the stellar halo from surveys pointed toward the Magellanic Clouds and the Andromeda galaxy.

Removing Atmospheric Fringes from Zwicky Transient Facility i-band Images using Principal Component Analysis

Here we present a method for constructing models of these atmospheric fringes using Principal Component Analysis that can be used to identify and remove these artifacts from contaminated images. In addition, we present the Uniform Background Indicator as a quantitative measurement of the reduced correlated background noise and photometric error present after removing fringes. We conclude by evaluating the effect of our method on measuring faint sources through the injection and recovery of artificial stars in both single-image epochs and co-additions. Our method for constructing atmospheric fringe models and applying those models to produce cleaned images is available for public download in the open source Python package fringez.

Gravitational Microlensing Event Statistics for the Zwicky Transient Facility

We predict that the Zwicky Transient Facility (ZTF), an all-sky optical synoptic survey that observes the entire visible northern sky every few nights, will observe ∼1100 microlensing events in 3 yr of observing within 10° latitude of the Galactic plane, with ∼500 events in the outer Galaxy (ℓ ≥ 10°). Using the microlensing modeling software PopSyCLE, we compare the microlensing populations in the Galactic bulge and the outer Galaxy. We also present an analysis of the microlensing event ZTF18abhxjmj to demonstrate how to leverage these population statistics in event modeling.

Transformations from Pan-STARRS1 and UBV Filters into ZTF Filters

The Zwicky Transient Facility (ZTF) is an optical time-domain survey that surveys the visible Northern sky every few nights in ZTF g-band and r-band, as well as an i-band filter used only for partnership observations and thus absent from this analysis. Transformations between the photometric systems of ZTF, Pan-STARRS1 (PS1), and Johnson–Morgan–Cousins (UBV) are essential for extending both the time-baseline and wavelength coverage of ZTF catalogs with information from other catalogs. We use cross-matching catalogs and simple stellar population synthesis models to derive photometric transformations from PS1 and UBV to ZTF.

Accomplishments

Actor
September 2007 - June 2015

Over my four years living in Chicago I performed in nine shows, acted in five films, assistant directed three productions, and taken over a dozen classes in acting for the stage, film, improv and Meisner technique. In 2012, along with several other theatre artists, I founded The Poor Theatre as a way to build such an ensemble. We strove to explore those parts of the human experience often overlooked. We worked to give voice to stories that often go unheard. My contributions to the company were both as a member of the acting ensemble and the Managing Director, before I left the company in 2015.

Physics Teacher
July 2011 - June 2015

I tutored high school, undergraduate, and post-baccalaureate students in physics and astronomy in my years between Northwestern and UC Berkeley. My teaching style is greatly informed by my time spent as an actor, where I learned that communication is a skill one develops and hones with practice. My time has also been spent working to develop physics and robotics programs for middle and high school students in Evanston, IL (Northwestern University Center for Talent Development) and Oak Park, IL (Oak Park Education Foundation). Using all facets of STEM curriculum, I worked with students to analytically, strategically and cooperatively design, engineer, build, program, test and operate competition robots.

Teacher

Astrophysicist

Actor

Father