Shailen Sheth — Finance & Ops Builder

What I build outside work

Because builders don't clock off

Ledger

Finance careers at companies worth working for · Work in progress

The curated job board for finance professionals in tech & AI. LinkedIn buries signal — Ledger cuts through it. Curated roles only, AI-powered CV match scores, a company explorer map, and a live funding radar so you know who's about to hire before they post the role.

Product design AI match engine Company map Funding radar B2B SaaS

247

Roles live

1,200

Beta candidates

Paying companies

140+

Companies mapped

Pitch deck — click any slide to expand

Deal Gatekeeper

Pre-sync CRM-to-ERP validation agent

An AI agent that sits between closed-won CRM deals and the financial system. Before any deal syncs, it runs through data quality, pricing reconciliation, deal type classification, and revenue recognition checks — then outputs a structured decision with specific questions for the CFO. Built to eliminate bad data and commercial rule violations from contaminating the books.

AI agent Structured output Rev rec CRM-to-ERP Prompt engineering

Decision paths

Validation layers

Silent failures

system_prompt_v1.0.txt CONFIDENTIAL

// ROLE You are a Pre-Sync Deal Gatekeeper. You analyse closed-won CRM deals before they sync to the ERP. Your role is to prevent bad data and commercial rule violations from contaminating the financial system. // YOUR DECISION For every deal, output exactly one of three decisions: APPROVE Deal passes all checks. Sync as-is. AUTO_CORRECT Deal has a non-commercial data quality issue that can be safely fixed. Apply the correction, log it, then approve. Auto-correct is permitted ONLY for: (a) missing currency where territory unambiguously indicates one (b) reformatting amount fields (e.g. "300,000" → 300000) (c) trivial typo corrections in non-commercial fields Auto-correct is NEVER permitted for: pricing structure, deal type classification, deal owner, legal entity, or anything affecting commission or revenue recognition. When in doubt, escalate to FLAG instead. FLAG Deal has an issue that requires human judgment. Do not sync. Output specific, actionable questions for the CFO. Never output "please review" — always specify the decision required. // PRICING RULES Standard Platform: $150,000 platform fee + $150,000 per Digital Worker. Minimum 1 Worker. Entry price: $300,000. Enterprise Platform: $500,000 platform fee + $125,000 per Digital Worker. Minimum 5 Workers. Entry price: $1,125,000. Multi-year uplift: Standard is 5% compounding annual uplift in Year 2+. Non-standard uplifts must be flagged. Legacy SKU: No new deals permitted. FLAG any new deals using the legacy SKU. // REASONING PROCESS (ordered) 1. DATA QUALITY CHECK Are all required fields present? Is close date in the past? Is ARR a number? Is currency populated? 2. PRICING RECONCILIATION Does the stated ARR reconcile to a valid pricing configuration? Show the math explicitly. If ARR matches a config that violates a minimum, call it "non-compliant" — never silently pass it. 3. DEAL TYPE VALIDATION For expansions: does the legal entity match the existing contract holder? If not → new logo. Wrong deal type affects commission. 4. REV REC FLAGS Scan for: acceptance clauses, bundled services, non-standard ramp structures. // IMPORTANT BEHAVIOURS // When in doubt, FLAG. Never auto-correct anything affecting pricing, // deal type, owner, or entity. // Show your math. The CFO will check it. // Never invent data. If a field is missing, say so. // Questions must be answerable — "please review" is not acceptable.

Three possible decisions

✓ Approve

Sync as-is

Deal passes all four checks cleanly. No issues found. Syncs to ERP immediately with no human intervention.

⚡ Auto-correct

Fix, log, then approve

A non-commercial data quality issue can be safely resolved automatically — formatting, missing currency where territory makes it unambiguous, trivial typos. Correction is logged. Never applied to pricing, deal type, entity, or anything affecting commission.

⚑ Flag

Hold for CFO decision

Anything requiring human judgment. Deal does not sync. Output is specific, decision-ready questions — not "please review." The CFO gets exactly what they need to resolve it.

Four-step reasoning process (always in this order)

Data quality

All required fields present? Close date in the past? ARR is a number? Currency populated? Missing fields → FLAG, except currency which can auto-correct.

Pricing reconciliation

Does stated ARR match a valid configuration? Math is shown explicitly. If it matches a config that violates a tier minimum — that's non-compliant, not a pass.

Deal type validation

Expansion deals: does the signing entity match the existing contract holder? If not, it's a new logo. Wrong classification affects commission treatment.

Rev rec flags

Scan for acceptance clauses, bundled services, non-standard ramp structures. These don't block sync but must be flagged for finance review.

⚑ FLAG — Deal held for CFO review

A deal comes in classified as an Expansion at $875,000. The pricing maps to Enterprise + 3 Digital Workers — but Enterprise requires a minimum of 5. Additionally, the signing entity differs from the existing contract holder. Two separate issues, one decision.

Input — CRM deal record

"deal_id": "DL-2026-0441",
"deal_owner": "Sarah Mitchell",
"customer_name": "Meridian Logistics Group",
"signing_entity": "Meridian Logistics GmbH",
"existing_contract_holder": "Meridian Logistics Ltd",
"deal_type": "Expansion",
"platform_tier": "Enterprise",
"digital_workers": 3,
"arr": 875000,
"currency": "USD",
"close_date": "2026-04-22",
"contract_years": 1,
"ae_notes": "Customer expanding EU ops.
  New German entity signing."

Output — Gatekeeper decision

{
  "deal_id": "DL-2026-0441",
  "decision": "FLAG",
  "summary": "Two issues block sync: Enterprise
    pricing with only 3 DW violates the
    5 DW minimum, and the signing entity
    (GmbH) differs from the existing
    contract holder (Ltd), making this
    a potential new logo.",

  "data_quality": {
    "status": "PASS",
    "issues": []
  },

  "pricing_validation": {
    "status": "NON_COMPLIANT_CONFIG",
    "stated_arr": 875000,
    "expected_arr_for_stated_config": 875000,
    "math": "Enterprise: $500K + (3 × $125K)
      = $875K. Math reconciles, but
      Enterprise requires min 5 DW.
      Compliant Enterprise + 5 DW
      = $500K + (5 × $125K) = $1,125K.",
    "issue": "Config violates 5 DW minimum.
      Nearest compliant tiers:
      Standard + 5 DW = $900K, or
      Enterprise + 5 DW = $1,125K."
  },

  "deal_type_validation": {
    "stated_type": "Expansion",
    "correct_type": "Unclear — requires CFO decision",
    "reason": "Signing entity (Meridian GmbH)
      differs from contract holder
      (Meridian Ltd). Per policy, entity
      mismatch = treat as new logo.
      Affects commission calculation."
  },

  "rev_rec_flags": [],
  "auto_corrections_applied": [],

  "human_review_questions": [
    "1. Should this be classified as an
    Expansion (same commercial group,
    different legal entity) or a New
    Logo (different entity = new
    contract)? This determines
    commission treatment.",
    "2. The stated ARR of $875K maps to
    Enterprise + 3 DW, which violates
    the 5 DW minimum. Was a pricing
    exception approved? If not, which
    compliant tier applies: Standard
    + 5 DW ($900K) or Enterprise
    + 5 DW ($1,125K)?"
  ],

  "ready_for_light": false
}

More coming

Next project in progress. This page grows as I build.

Oct 2023 — Present

Paddle

Accounting Manager

Built finance infrastructure at a global payments platform from the inside. Implemented NetSuite from scratch, automated reconciliations and reporting using AI tooling, and cut close cycle time by 40%. Finance ops, compliance, and cross-team process work across a fast-moving, multi-entity SaaS business — a lot of it without a playbook.

NetSuiteIFRSAI workflowsFinOpsMulti-entityProcess automation

2022

APEXX Fintech

Financial Accountant

Finance and compliance operations across a high-growth payments startup in full build mode. Fast-paced, multi-jurisdictional, and light on process — a lot of it had to be built or figured out on the fly. Got things done without a playbook.

FintechPaymentsMulti-jurisdictionComplianceStartup finance

2015 — 2021

DSK Partners

Trainee → ACA Qualified

Trained ACA with ICAEW across multiple client sectors — audit, tax, financial reporting, legal entity management, and advisory work. Led a NetSuite implementation from scratch for a key client. The breadth-first foundation that means I can operate across finance, legal, compliance, and ops without needing a specialist for every sub-function.

ACA qualificationMulti-sectorNetSuiteLegal entitiesUK GAAPClient advisory

Shailen
Sheth

Proof over promises

Because builders don't clock off

Built for the messy, moving, from-scratch stage

The toolkit

Ready to build what doesn't exist yet

If you need someone to build it — not manage it

ShailenSheth

Proof over promises

Because builders don't clock off

Built for the messy, moving, from-scratch stage

The toolkit

Ready to build what doesn't exist yet

If you need someone to build it — not manage it

Shailen
Sheth