Skip to content

Two ways in, both fixed-price

Most engagements start with the audit, because it is the cheapest way for both of us to find out whether I can help. Prices are below rather than behind a call.

Available for new work

Based in Pune, India (+05:30). Local time —.

Your time
—
Working-hours overlap
—

I hold afternoons and evenings IST open for calls, which covers European mornings and US mornings on the earlier side. Async-first either way.

Most people start here

Fixed scope · 1–2 weeks

Agent Production Readiness Audit

$4,000 – $6,000

A fixed-price diagnostic on an agent you already have running. You keep everything I build.

What you get

  • Failure taxonomy for your specific agent, ranked by production risk
  • Cost and latency profile — where tokens actually go, and p50/p95 per operation
  • An eval suite wired to your real traffic that your team owns and keeps running
  • Prioritised fix list with effort estimates, sequenced by risk reduction per day of work
  • A 90-minute walkthrough with your engineers, recorded

Right for you if: You have an agent in or near production and a growing feeling that you cannot prove it works.

Enquire about the audit

Monthly retainer · ~2 days/week, 3-month minimum

Fractional AI Reliability Lead

$5,000 – $8,000 / month

I own agent reliability as a function: evals, observability, and cost discipline, alongside your team.

What you get

  • Own and grow the eval suite as your agent surface expands
  • Tracing and alerting on the failure modes that actually cost you money
  • Model routing and token budgets — usually the fastest measurable win
  • Design review on new agent features before they ship, not after
  • Direct line to me on Slack, and a written monthly reliability report

Right for you if: Agents are core to your product and nobody currently owns whether they work.

Enquire about the retainer

Ranges reflect scope, not negotiation. A single agent with existing traces sits at the lower end; several agents with no observability at all sits at the upper end. You get the exact figure in writing before anything starts.

How an engagement actually runs

No discovery phase billed as a project. No proposal deck. Five steps, and you can stop after any of them.

  1. 1

    Message

    Tell me what your agent does, roughly what volume it handles, and what worries you. Two paragraphs is plenty. I reply with an honest read on whether I can help — including when I think you do not need me.

    Same day, usually
  2. 2

    A 20-minute call

    Not a sales call. I ask about your architecture, your traffic, and what has already gone wrong, so I can scope accurately. You get useful answers whether or not we work together.

    20 minutes
  3. 3

    A written scope

    One page: what I will look at, what you will get, the fixed price, and the dates. No hourly estimates that drift, no discovery phase billed separately.

    Within two days
  4. 4

    The work

    Read access to the code, whatever traces exist, and 60–90 minutes with an engineer who knows the system. I work async and share findings as I go rather than saving them for a reveal.

    1–2 weeks for an audit
  5. 5

    Handover

    A recorded walkthrough with your engineers, plus everything I built — the eval suite, the traces, the fix list. It is yours whether we continue or not.

    90 minutes

What I will not do

Stated up front, because a consultant who claims to do everything is telling you they specialise in nothing.

Build your agent from scratch

I am useful once something exists and needs to survive contact with users. If you are at the blank-page stage, you need a product decision, not a reliability engineer.

Fine-tune or train models

Almost every problem sold as needing a fine-tune is a retrieval, evaluation, or prompt-architecture problem. I will tell you that for free, and if you genuinely need a fine-tune I am not your best option.

Staff augmentation by the hour

Hourly billing rewards slowness and invites geographic rate haggling. I quote fixed scope so we both care about the outcome instead of the timesheet.

Sell you an audit you do not need

If the free scorecard shows you are already in good shape, I will say so. Selling an audit to a team that does not need one costs me a reference, which is worth more than one invoice.

Questions about working together

Why are prices on the page when most consultants hide them?

Because hiding them wastes both our time. If $5,000 is outside your budget you should know that in ten seconds, not after two calls.

It also signals that the price is the price. Quoting after discovery invites the suspicion that the number depends on what you can afford.

Do you charge less because you are based in India?

No. The price is the same regardless of where the client or I happen to sit, which is exactly why I quote fixed scope rather than an hourly rate.

You are buying a specific outcome — an eval suite, a cost profile, a ranked fix list — and that outcome has the same value to you whoever produced it.

What if the audit finds nothing serious?

Then you get a short report saying so, and I will tell you honestly that you do not need a retainer. That has happened, and those teams have sent me referrals.

In practice it is rare: most teams running agents in production have at least one critical gap, most often no tool-argument validation or no cost ceiling.

Can we start with something smaller than an audit?

Yes. Send me your scorecard result or a description of one specific failure, and I will give you my read at no charge. I would rather be useful early than gate everything behind an invoice.

If that turns into a paid engagement, good. If it does not, you still got an answer.

How do you handle our code and data confidentially?

Happy to sign your NDA before anything is shared. I work on the minimum access needed — read access to the relevant repositories rather than broad admin, and anonymised or synthetic samples where real data is not necessary.

I do not need production credentials to audit an agent, and I will push back if offered them.

Are you available for a full-time role instead?

I will listen to the right one. I am doing independent work deliberately, but a genuinely interesting remote role with location-agnostic compensation is worth a conversation.

Message me with the company, the role, and the comp range and I will be straight with you either way.

Start with a message, not a commitment.

Describe your agent and what worries you. I will tell you what I would look at first and whether it is worth paying me — that answer is free, and sometimes it is no.

Message me on WhatsAppBook a 20-minute callor email instead

Not sure which you need?

Run the free scorecard first. It takes four minutes and will tell you whether your gaps are the kind a one-off audit closes or the kind that needs someone to own them.

Run the scorecard