About the role

🚀 Who we are & what we're building

Peakflo is a rapidly growing Agentic AI company. We are revolutionising the way global finance teams work with agentic workflows, and we are actively seeking exceptional talent across diverse disciplines to champion this transformation.

Our growth story: Peakflo is backed by top-tier global accelerators and investors. We are proud alumni of Y Combinator (W22) and the Google AI Accelerator. Our momentum has been recognised globally by top tech and finance publications:

• TechCrunch: Peakflo's bid to build business payments for Southeast Asia attracts capital, customers
• PYMNTS: Peakflo raises $4.1M in seed funding to streamline vendor payments
• Grit Daily: Latest feature on Peakflo's innovations

Our culture: We believe in building a vibrant, high-performance culture that rewards curiosity, ownership and innovation. Our team spans the globe, and we love coming together to solve hard problems and celebrate our wins. ❤️

Who you'll learn from: You'll work directly with our founders, including a former McKinsey engagement lead who went on to found and scale a Y Combinator company. If you're coming from consulting, this is the rare startup where the structured-thinking muscle you built is the point of the hire rather than something to be quietly tolerated.

Concretely, here is what that access buys you in your first year. We'll hold ourselves to it:

• Weekly working time with a founder on live accounts, not a monthly skip-level.
• You present delivery health to the company. Not a slide someone else assembles from your input. Your read, your recommendation.
• You own a function, not a workstream. Within 12 months you should be hiring into it. That is a P&L-adjacent leadership track, on a timeline no consultancy pyramid will match.

🎯 Why this role exists

Most job descriptions describe a function that already works and needs another pair of hands. This one describes a function you will be building.

Peakflo is winning enterprise deals faster than its delivery and implementation team has scaled. Our implementations involve multi-entity groups running SAP/Oracle/MSD365, third-party middleware built by the customer's own IT team, phased UAT across a dozen-plus client stakeholders, and integration surfaces (AI-OCR extraction, PO matching, approval policies, ERP posting) where one configuration change can ripple across a hundred test cases. Until now we have delivered these founder-led and engineer-led: on judgement, product depth and effort rather than on a system. That got us here. It does not get us to ten concurrent enterprise implementations.

What we're missing is the connective tissue. A single source of truth every stakeholder trusts. Test evidence captured at the point of testing rather than reconstructed afterwards. Dependencies on the customer's own IT teams chased as hard as our own. Dates that move loudly and early instead of quietly and late. The discipline that turns a three-hour client workshop into tracked, owned, dated work, and keeps it honest in week nine as reliably as in week one.

To be precise about the ask: this is not a resourcing problem. You'll have engineers, QA, ML engineers and forward-deployed engineers to draw on, and product and engineering leadership are a message away. Capacity has not been our constraint. Orchestration has.

So you are not inheriting a mature delivery function to maintain. You are being hired to design one, prove it on our most complex live account, and make it the default way Peakflo delivers every enterprise implementation after that.

If that's the problem you want, we should talk. If you'd rather join somewhere the operating model already exists, this genuinely isn't the role.

💪 What you'll own

You'll carry three to four enterprise accounts at once, deliberately staggered across the implementation lifecycle: AS-IS discovery, TO-BE design, configuration, SIT, UAT, go-live, hypercare. The back half of that sequence, from SIT onwards, is far heavier than the front, so a portfolio only works if the phases are offset. Accounts roll off after hypercare with a clean handover to a CSM, and a new one comes on. Reading your own capacity honestly, and saying you're at the limit before you're underwater rather than after, is part of the job rather than an admission of failure.

1. The single source of truth. You own the RAID log, the UAT test tracker and the delivery plan for your accounts, and you own the discipline that keeps them true. Every issue, risk, action, decision, dependency and new requirement that surfaces, whether on a call, in a chat or in an email, lands in the tracker the same day with a type, a severity, an owner and a date. Including the ones where the answer is "no action needed": a closed decision is a logged decision.

2. The timeline, and the honesty of it. You own the master plan for each account: phases, milestones, workstream dependencies and a named critical path, baselined and then tracked against actuals. You know at any moment which milestone is at risk and why, and you can say what would have to be true to recover it. Two things matter more than the plan's elegance. First, you forecast slippage rather than report it: a date that moves should have been visible to you weeks earlier in test-execution burn rate, defect-closure rate or an unanswered client dependency, and flagged then. Second, the plan reflects reality even when reality is unwelcome. We would far rather carry an amber milestone with a recovery option than a green one that turns red the week before go-live. You also own capacity planning against that timeline: what has to be true about pod availability, client tester availability and third-party integrator bandwidth for each milestone to land.

3. UAT and test governance. You run structured UAT with enterprise finance teams: test scripts derived from the signed-off design document and written before the session, cases assigned to named testers, verdicts recorded with the tester, date and document reference at the moment of testing. You define and enforce regression rules, so that when an extraction, matching or posting rule changes, the agreed regression set is re-run and the run is evidenced, not asserted. UAT weeks are also when you travel: sitting next to the client's testers for a week surfaces more, and settles more, than a month of calls.

4. Client stakeholder management and escalation. You are the operating counterpart to the customer's finance controller, IT lead and project sponsor. You chase client-side dependencies as hard as internal ones. You bring bad news early, with a revised date and a mitigation, rather than late with an explanation. You know the difference between a client request that is in scope, one that is a change request, and one that should be politely refused. And you can say the third one out loud.

5. Engineering handover and throughput. Every tracked item requiring engineering becomes a ticket the same day, linked from the tracker row, with enough written context that an engineer doesn't need to attend the call. You then keep those tickets moving, flagging stalls, unsprinted priority work, and tickets whose status no longer matches reality, and you route them to the right function rather than escalating everything to engineering.

6. The implementation lifecycle, end to end. You carry each account from discovery to roll-off, but "own" means something different at each stage and the distinction matters. Through AS-IS and TO-BE, the product manager owns the design; you don't write it. You make sure it lands: open items tracked, minutes issued after every session, action items chased on a cadence, and the design signed off in writing by the client before anything gets built. What you do owe is comprehension. The UAT script is built from that design, and nobody can write test scripts for a design they only half-followed. Through configuration, you hold the product team to proper scoping: tickets scoped, assigned to a sprint, tracking against their internal test dates, with misses escalated early rather than absorbed quietly. Every configuration is documented, and internal testing runs against the signed-off design document rather than against somebody's recollection of the workshop. SIT, UAT, go-live and hypercare you run directly. You define what "ready" means before cutover, covering data migration, vendor master, approval policies, integration cutover and user training, you run the cutover and stay through stabilisation, then close out with a real knowledge transfer to the CSM who takes the account on. One rule throughout: signoff is written, never implied. A TO-BE design nobody countersigned is a change request waiting to happen, usually at the worst possible moment.

7. Executive reporting. A weekly status a CFO can absorb in two minutes: what moved, what's blocked, what's at risk, what needs them. No status theatre, no green projects that turn red overnight.

8. Directing the delivery pod. You won't do this alone and you're not expected to. You'll have engineers, QA, ML engineers and forward-deployed engineers to task, and directing that capacity well is a core part of the job, not an aside. Deciding what gets tested by whom, what an FDE picks up on the client's environment versus what goes into the product backlog, and where MLE time is worth spending on an extraction problem: those calls are yours. You may delegate the work of maintaining the trackers. You cannot delegate whether they are true.

9. Turning delivery signal into product. You sit closer to enterprise pain than anyone else in the company. You translate recurring implementation friction into clear, prioritised product input, with evidence of frequency and revenue impact rather than anecdote. This is a real and valued part of the role; it is not the majority of it.

🧰 What you'll have

We'd rather over-communicate this, because it shapes what we expect of you:

• A delivery pod to direct: engineers, QA, ML engineers and forward-deployed engineers you can task against implementation work.
• Direct access to product and engineering leadership, with no ticket-shaped queue between you and a decision.
• Founder air cover for the hard conversations: scope pushback, date resets, saying no to a customer.
• Authority to set the operating model, not just to follow one. The standards below are the outcome we want; how you get there is yours to design.

The corollary is real accountability. With that support in place, a client commitment that quietly slips, or an issue raised on a call that never becomes a tracked row, is an orchestration failure, and it's the failure this role exists to prevent.

📏 The operating standards you'll be held to

We're publishing these because they're the substance of the job and we'd rather you self-select. These are the standards, not aspirations:

• Same-day logging. Anything raised on a client call is in the tracker before you close your laptop. "Discussed, no action agreed" is a logged outcome with a reason, not a reason to skip logging.
• Evidence over assertion. Any claim that something is tracked, tested, fixed or delivered can be answered immediately with a row number, ticket ID, test case or link.
• Same-day engineering handover. Items needing engineering become tickets the same day, linked from the tracker row.
• Test evidence at source. Every verdict carries tester, date and document or reference number, recorded when the test runs, never reconstructed later from chat logs.
• Regression discipline. Any change to an extraction, matching or posting rule triggers the agreed regression set. The run is evidenced with scope and result.
• Dependencies are first-class. Client-side and third-party items are tracked with a named owner on their side, a date, and a chase cadence. "Waiting on the customer" is a tracked state, not an excuse.
• The plan is baselined and current. Every account has a milestone plan with a named critical path, baselined and tracked against actuals. Status is reported against the baseline, not against last week's revised guess.
• Slippage is forecast, not reported. Risk to a milestone is raised when the leading indicators move (burn rate, defect closure, an unanswered dependency), not when the date arrives. An amber milestone with a recovery plan beats a green one that turns red late.
• Dates are communicated before they slip. If a commitment is going to move, the customer hears it from you before the deadline, with a revised date.
• Ownership rules are written, not folk knowledge. Conventions about what needs a ticket, who owns what and when something is exempt are documented and applied consistently, not held in one person's head.
• Coverage is arranged, not assumed. Planned absence comes with a written handover and a named cover for live accounts.

🗓️ Your first 90 days

• Days 1–30: Own the delivery cadence on our most complex live enterprise implementation. Audit the current RAID log and test tracker against call recordings and chat history; produce a reconciled backlog of everything that was discussed but never logged. Publish the first weekly executive status.
• Days 31–60: Stand up the operating model: tracker schema, severity and ownership definitions, regression policy, UAT script standards, escalation matrix, status template. Publish a baselined milestone plan with a named critical path, agreed with engineering, product and the client. Close the reconciled backlog.
• Days 61–90: Prove it holds without heroics. Zero same-day logging misses for a full month. A go-live readiness plan the customer has signed. A written playbook another delivery manager could pick up and run, because the second and third enterprise accounts are coming.

🕵️ Who we're looking for

Experience: three routes in, weighted equally.

• Around 5 years in enterprise ERP delivery at a consultancy or systems integrator (Accenture, Deloitte, PwC, EY, KPMG, Capgemini, LTIMindtree, Infosys, TCS, Cognizant or similar), where you led a workstream on the client side rather than only a technical stream. An SAP FI/CO, MM or P2P functional background is a strong plus: knowing what a company code, a three-way match or a GRN actually is will save you months here.
• Around 5 years leading enterprise B2B SaaS implementations, where you personally owned UAT, defect triage, integration testing and cutover on ERP-integrated go-lives.
• Around 5 years in strategy consulting followed by an operating or delivery role. We're interested in the structure that training gives you, but we'll want evidence you've since shipped software rather than only advised on it.

We care far more about what you've personally run than where you ran it. Someone from a systems integrator who has taken an SAP-integrated go-live through UAT and cutover is a stronger candidate for this job than a strategist with a better-known logo. Please don't self-reject on brand, and don't self-reject on seniority either: if you've been running the workstream while someone more senior presented it, this is the role where you stop handing over the microphone.

Non-negotiables:

• You have owned a RAID log, issue register or equivalent as a live working instrument, and you can still work in one yourself. Delegating its upkeep is fine; being unable to answer "which row?" on the spot is not.
• You have owned the plan for a workstream or a programme with a real critical path, and have called a slip weeks ahead of the date rather than on it. Be ready to talk us through one you got wrong, and what the leading indicator was that you missed.
• You can direct a mixed technical pod (engineers, QA, ML and forward-deployed engineers) and get committed dates out of people who don't report to you.
• You have run SIT and UAT with enterprise customers: written the scripts, assigned the cases, chased the testers, defended the exit criteria.
• You are comfortable in the technical weeds. You can read an API payload, reason about a field mapping, tell a real defect from a configuration gap, and challenge an engineer's estimate without needing a translator.
• Exceptional written English. You can compress a three-hour workshop into eight lines a CFO will act on.
• You have told a customer something slipped, to their face, with a plan, and kept the relationship. Once is enough. We care that you've done it, not how often.
• Extreme attention to detail, and the temperament to keep applying it on day 200 when nobody is watching.

Working pattern: India-based, remote. Working hours are 08:00 to 18:00 IST, with an hour for lunch. That keeps you live from 10:30 SGT through to the close of the Singapore working day, which is when our enterprise customers actually need us. Enterprise implementations run on the customer's clock, not ours. Client-site travel is light, around two weeks a year, and it lands on UAT weeks.

➕ We're particularly interested in people who have

1. Delivered into large-enterprise ERP environments such as SAP (S/4HANA or ECC), Oracle Fusion / E-Business Suite, Microsoft Dynamics 365 F&O, Infor or Workday Financials, and understand how APIs, SFTP and customer-built middleware actually behave in production. Multi-entity, multi-currency, multi-company-code estates with a systems integrator in the mix, not single-entity accounting packages.
2. A background in FinTech, accounting operations, CFO-stack software or supply chain / logistics, or the curiosity to get there fast. Knowing what a three-way match, a GRN or a credit note actually is will save you months.
3. Worked an implementation where the customer's own IT team or systems integrator sat on the critical path, and found a way to hold them accountable without owning them.
4. A working understanding of LLM capabilities, AI agents and their failure modes, enough to set honest expectations with a customer about what an agen