MAXIMAND

Careers

Cut the cost of AI in finance, and raise what it returns.

Maximand is the specialist advisory for financial institutions running large generative-AI workloads. We cut what banks, insurers, and funds spend on AI while raising the quality of the work and, for research teams, the strength of the signal. Every result is proven on the client's own invoices. We are growing, and hiring across engineering, client delivery, and measurement.

See the rolesWhat we believe

What you would be working on.

Most financial institutions cannot say what their AI spend is buying. We go through the usage, cut the spend that returns nothing, and hold or raise the quality of what remains. On a research desk that shows up as a lower cost per validated signal and more strength in the signal itself. In a bank or an insurer it shows up as the same work done for far less, with risk and compliance signing off the quality floor. We prove every result on the client's own invoices before any fee is charged.

The engagements are real and paid, and the method is published in full as the Token-Efficiency Standard, so the bar is in the open and we are held to it. You will measure real money, you will work from a method you can read, and you will sign your name to the number.

Open roles

The roles.

Engagement Principal, Financial Institutions

Client delivery · Remote, with client travel

You lead our engagements inside banks, insurers, and funds, from the first conversation to the signed result, and you earn trust in rooms full of senior, skeptical people: CFOs, security officers, and model-risk teams.

What you will do

  • Lead the Token Audit end to end at a financial institution, from intake to a costed roadmap.
  • Co-define the baseline and the quality floor in writing with the client's finance and risk functions.
  • Present a result finance will sign, cost down with quality and signal held or raised, and stand behind it.
  • Run the verification behind the gainshare, period by period, on the client's own ledger.

What we look for

  • A track record leading advisory, FinOps, or transformation work inside or alongside banks, insurers, or funds.
  • You can hold a room of senior, skeptical people and leave with their trust.
  • You write clearly and you are rigorous about evidence; you do not round in your own favor.
  • You are comfortable owning a number that has real money attached to it.

Token-Efficiency Engineer

Engineering · Remote, with client travel

You go into a financial institution's AI estate and take cost out without giving up quality, and often while raising it. You turn what works into tooling the next engagement reuses.

What you will do

  • Instrument a client's generative-AI usage and measure unit cost per task, request, or output, alongside the quality of what it produces.
  • Implement the levers: eval-gated cascade routing, prefix and semantic caching, model right-sizing, batching, distillation, and retrieval sizing.
  • Build the evaluation harnesses that hold a quality floor, and lift it where the work allows, as the bill comes down.
  • Turn each engagement into reusable measurement and tooling, so the method compounds.

What we look for

  • Production experience with LLM inference and the cost, latency, and quality tradeoffs between models.
  • You read provider pricing the way others read a balance sheet, and you know where the tokens actually go.
  • You can work inside a bank or fund's security and model-risk limits.
  • Pragmatic engineering: you ship the fix that moves the number, not the most interesting one.

Evaluation and Verification Lead

Research quality and signal · Remote

You own the measurement that proves both halves of our promise: that cost fell, and that quality and signal strength held or improved. For research teams that means a defensible cost per validated signal. You build the evidence, and you keep us honest.

What you will do

  • Design how results are measured: unit cost per task, A/B holdouts, frozen run-rate baselines, and the quality and signal metrics that sit beside them.
  • Build the evaluations that prove model quality holds, and that research signal strengthens, as cost falls.
  • Own the Engagement Ledger, so every engagement calibrates our model and adds to the benchmark.
  • Contribute to the open Token-Efficiency Standard and its falsifiable claims.

What we look for

  • A background in applied machine learning, evaluation, or experiment design, with real statistical rigor.
  • A low tolerance for claims that cannot be tested, including our own.
  • You care about being right in public, and you can defend a method to a quant.
  • You can turn messy production data into a measurement someone will sign off on.

How to apply.

Write to hello@maximand.ai with the role in the subject line. Skip the cover letter. Tell us about one time you cut a real bill or held a quality bar under pressure: what you measured, and how you knew it was true. If you are not sure you fit but you think you should, write anyway and say why.

Do not see your role here? If you are exceptional at something adjacent to this work, tell us what we are missing.