Part-Time Founding Engineer — Platform, Architecture, ML, Scaling, Team Building

We're seeking a high-hustle, risk-embracing, initially part-time founding engineer who sees the immense potential of what we've built and would be excited to get in on the ground floor. This job will start relatively small, and grow into a full-time one at the top of an organization. We need someone ready to take what we've built, scale it, and build a team around it as the job transitions to full-time down the road--whether that's in a few months or a year.

What do we do? Something no one ever has, and that could change business. Watch this.

This is the result of 15 years of research, testing on over 3bn records, and peer review. It's real. We have it up and running, with a containerized LLM-wrapped python core. We have a front end that is not yet connected to it. We have a paying client and a crap-ton of data. We have an ambitious GTM plan. To execute on it, we have a lot we need to build, deploy and support.

Engagement: Part-time to start (scoped equity-based milestone contract), scaling directly to Full-Time Founding Engineer as sales pipeline/strategic fundraising materialize.
Location: Los Angeles, CA preferred (PST or other US remote acceptable).
Reporting Structure: Reports directly to the CEO/Founder.
Ideal Background: 5+ years shipping production-grade Machine Learning Operations (MLOps), distributed data platform engineering, or multi-tenant cloud-native architectures.

This is the first engineering hire, working directly with our founder — a USC professor whose research underlies the scoring methodology. You'll take a system that already computes influence scores from trade graphs and turn it into something that runs reliably in a stranger's cluster, with the modeling judgment to know when the counterfactual model itself needs to change, not just the code around it.

What you'd own
The multi-cloud platform — the Terragrunt/AKS stack that runs Metaflow + Argo Workflows today, with multi-cloud portability (AWS/EKS and Google Cloud/GKE) as a primary requirement: node pool sizing, cost (spot vs. on-demand), the drift-detection and CI/CD pipeline, and the security backlog (Defender for Cloud, Azure Policy, CSI Secrets Store, image scanning).

The modeling roadmap — deciding what gets built next: new data sources beyond the current DEX/CEX pipelines, edge-pruning and matrix operations on user transaction graphs scaling to hundreds of millions of entities and billions of edges, and when a client's data volume justifies moving from single-node Pandas to Dask or Spark.

Client on-prem deployments — packaging the same container that runs on our AKS cluster so it deploys clean on a client's own Kubernetes (EKS, GKE, or bare on-prem), with no data leaving their network by default. You're the person on the call when their infra team asks "where does our data go."

The build-vs-buy calls — REST API for scoring-as-a-service, schema versioning across client installs, which storage backends (S3, GCS) actually get built next. You set the backlog, not just clear it.

The offshore team's technical direction — scoping their sprints, reviewing their infra and backend PRs, deciding what's safe to delegate versus what you do yourself.

Where the codebase is today
Two repos, both real and running: SocialValue Infra (Terragrunt/Terraform, deploys AKS + Blob + Postgres + Key Vault + Argo + ingress) and SocialValue Engine (Python/Metaflow flows — ingest → build features → score — scheduled daily via Argo cron). A Random Forest fits a counterfactual per asset; the delta between actual and no-neighbor-predicted outcome propagates along the graph as SocialValue.

Stack: Terragrunt/Terraform · AKS · Argo Workflows · Metaflow · Python/polars/scikit-learn · Azure Blob + ACR + Key Vault · Docker/Helm · Claude API (insights layer)

Who This Is For
An individual who has owned data pipelines end-to-end at scale and thrives in the highly autonomous environment of a pre-Series-A company. You possess deep systems chops in Kubernetes, Terraform, and cloud-native networking, combined with the modeling judgment to understand how algorithms behave when exposed to real-world transactional data. You want to build the architectural rails to commercialize rigorous, peer-reviewed methodology.

You are comfortable running an FDE operation, internal and/or outsourced.

You are not daunted by a startup and its risks, and you are willing to take a chance on something that may never work out. You may never make a dime, and you accept that risk because of the chance to get in on the ground floor of something that could be transformative.

You are comfortable starting with a part-time equity-only position and scaling to a full, higher-equity (and paid) one as the company grows and finds traction.

You have solid communication skills.

You pay attention to detail and can figure things out. Prove it in any contact by pointing out the flavor in the first sock monkey picture on my site, and by including three people who can serve as references and attest to your skills, experience, and character.

Not a Fit If
You expect a pre-scoped, neatly organized ticket queue. You will be the architect who builds the queue.

Your goal is to conduct novel academic ML research. The academic methodology is locked and peer-reviewed; your metrics of success are deployment velocity, data security, and platform stability at scale. R&D refinements will come later.

You rely on AI to do your thinking for you. Use it to take your ideas farther and faster, sure, but if you think handing in a generic AI-written report to the CEO is a good idea, this is the wrong place.