About Flowt

Flowt is building a financial intelligence platform to unlock access to financing for climate-smart small and growing businesses (SGBs) in Africa. We help SGBs become investor-ready by transforming bank and M-Pesa statements into cash-based profit & loss and cashflow statements and generating financial insights, then lending on the basis of this data.

Our platform combines:

• Structured onboarding and financial data capture
• Integrations with bookkeeping and banking systems
• Workflow automation for financing products such as invoice financing, purchase order financing, inventory financing, and asset financing
• AI-assisted validation, underwriting, and scoring systems to improve speed, accuracy, consistency, and fairness in credit decisions

We work closely with lenders and financing partners to help businesses access capital faster and more reliably.

About the Role

We are hiring a Data Analyst to sit at the centre of Flowt's data and AI ecosystem — the person whose judgment and attention to detail keeps every layer of the platform honest. As Flowt ingests financial data from multiple ERP and accounting systems, processes financial documents through AI extraction pipelines, calculates complex credit and financial metrics, and runs internal models to support credit decisions, this role ensures that what comes out of those systems is correct, complete, and trustworthy.

This is a human-in-the-loop role in the most meaningful sense: you are not just reviewing outputs — you are the feedback loop that makes the system smarter over time. Your corrections improve AI model accuracy. Your observations identify gaps in financial data quality. Your analysis of metric outputs flags formula errors before they reach a lender. You will work closely with the engineering, AI, and operations teams, translating what you see in the data into concrete improvements to the platform.

This role suits someone with a strong analytical foundation, genuine curiosity about financial data, and the discipline to be rigorous even when working under time pressure. You will handle real financial data for real businesses, and accuracy here has real consequences.

What You'll Own

1) ERP & Accounting Data Review

Flowt pulls financial data from multiple ERP and accounting platforms on behalf of borrowers. Your job is to ensure that what arrives is complete, correctly mapped, and ready to use:

• Review financial data synced from ERP and accounting integrations (Zoho Books, Xero, Odoo, QuickBooks, and others) for completeness, structural integrity, and logical consistency
• Validate that chart-of-accounts mapping has been applied correctly: that income, expense, asset, and liability accounts have been classified into the right categories and that unusual or uncategorised accounts are flagged for review
• Identify data quality issues in ERP-sourced records: duplicate transactions, missing periods, mismatched currencies, negative balances in unexpected accounts, and incomplete sync histories
• Cross-reference ERP data against other submitted financial documents (bank statements, audited accounts) to detect discrepancies — revenue figures that do not reconcile, expense categories that appear inconsistent, or balances that do not agree across sources
• Document recurring data quality patterns by ERP source and feed findings back to the engineering team to improve mapping rules, validation logic, and sync monitoring

2) Financial Metrics Validation

Flowt calculates a structured set of financial metrics from ingested data to assess credit readiness. You are responsible for ensuring those metrics are accurate and meaningful:

• Review calculated financial metrics for logical consistency and correctness: revenue trends, gross margin, EBITDA, operating cash flow, cash conversion cycle, burn rate, runway, inventory turnover, and receivables ageing, among others
• Validate metric outputs against the underlying source data — tracing a calculated figure back through the formula to the raw inputs and identifying where errors have been introduced
• Flag metrics that appear anomalous: values that fall outside expected ranges for the business type, sector, or size; trends that contradict other signals in the same dataset; or outputs that suggest a formula has been applied to incorrect or incomplete data
• Review multi-currency metric calculations to ensure that FX rate application is consistent, that the correct rate dates have been used, and that the disclosed assumptions match the actual calculation
• Maintain a live log of metric errors and edge cases observed in production, structured in a way that the engineering team can use to improve formula definitions, calculation versioning, and validation rules

3) Document Extraction Review & AI Output Validation

Flowt uses AI to extract structured data from financial documents — bank statements, audited accounts, invoices, purchase orders, tax certificates, and more. You are the human reviewer who ensures those extractions are correct before they are used:

• Review AI-extracted data from financial documents — field by field where necessary — comparing outputs against the source document to identify errors, omissions, hallucinations, and formatting failures
• Assess confidence scores and extraction quality across document types: identify patterns in where the AI performs well, where it struggles, and what document characteristics (quality, layout, language, format) drive poor performance
• Provide structured corrections through the validation queue: not just accepting or rejecting outputs, but annotating specific fields with the correct value and a reason code so that corrections feed directly into model improvement
• Review extracted financial tables from bank statements and audited accounts for structural completeness: correct column headers, properly split multi-period data, correctly identified opening and closing balances, and accurate transaction categorisation
• Flag systemic extraction failures — recurring errors on a specific document type, bank, or format — and escalate them to the AI team with enough structured detail to drive targeted model improvements

4) AI Model Performance Monitoring

As Flowt develops and deploys internal AI and ML models, this role tracks how well those models are performing against real-world data — and raises the alarm when performance degrades:

• Monitor extraction accuracy metrics across Flowt's document intelligence models on an ongoing basis: track precision, recall, and field-level accuracy by document type, data source, and model version over time
• Maintain the model performance dashboard: accuracy trends, error rate by category, volume processed, human correction rate, and the delta between model versions after each update
• Identify model drift: patterns where a model that previously performed well begins producing more errors — whether due to new document formats, new borrower types, data distribution shifts, or changes in input quality
• Review the output of transaction categorisation models: assess whether income and expense categories are being assigned correctly across different business types, flag systematic miscategorisations, and provide corrected labels with supporting rationale
• Analyse anomaly detection outputs: review flagged transactions, unusual financial patterns, and risk signals to assess whether the model is flagging genuine issues or generating noise — and feed that signal back into the model improvement cycle
• Contribute to model evaluation benchmarks: help define what "correct" looks like for new document types or new metric calculations, build labelled test sets from reviewed cases, and participate in structured evaluation of new model versions before they are promoted to production

5) Data Quality Operations & Continuous Improvement

Beyond individual case reviews, this role drives systemic improvements to the quality of data flowing through Flowt's platform:

• Maintain a structured issue log of data quality failures, extraction errors, metric anomalies, and model performance gaps — categorised, prioritised, and linked to the business impact of each issue type
• Produce regular data quality reports for the product and engineering teams: error volumes by category, resolution rates, time-to-resolution, and trend analysis showing whether quality is improving or deteriorating
• Identify and document recurring data quality patterns that should be addressed with automated validation rules rather than ongoing human review — and work with engineering to implement them
• Review the completeness and consistency of data entering the credit assessment process: ensure that every application reaching a lender has been through appropriate data review, and that gaps or anomalies are disclosed and resolved before submission
• Contribute to the development of internal data quality standards, validation checklists, and review protocols that can be used to onboard and train future team members

6) Analyst Support & Operational Review

This role also supports the broader operations team with data analysis and structured financial review:

• Support the credit and operations team with ad hoc financial data analysis: pulling, cleaning, and interpreting data to answer specific questions about a borrower's financial position
• Assist in preparing borrower financial profiles for analyst and lender review: summarising key metrics, flagging data quality concerns, and ensuring that the financial picture presented is accurate and complete
• Review submitted financial documents against Flowt's document standards: correct format, correct period, matching entity details, and compliance with product-specific document requirements
• Contribute to internal reporting: portfolio data quality dashboards, borrower data completeness rates, and operational metrics that help the team understand the health of the data flowing through the platform

What Success Looks Like (First 90 Days)

Weeks 1–2: Learn the Data

• Develop a working understanding of Flowt's data flows: how ERP data is ingested and mapped, how financial documents are processed, how metrics are calculated, and how AI outputs are generated and stored
• Complete supervised reviews of ERP data sets, extracted documents, and calculated metrics across each major financial document type
• Learn the validation queue: how to work through AI output reviews, how to structure corrections, and how findings are fed back into the model improvement process

Weeks 3–6: Independent Review & Pattern Recognition

• Independently validate ERP data, financial metrics, and document extractions within agreed turnaround times and accuracy standards
• Identify and document at least three recurring data quality or model performance patterns — with enough structured detail for the engineering or AI team to act on
• Begin contributing to the model performance log: tracking accuracy trends, flagging degradation, and annotating edge cases with corrected labels

Weeks 7–12: Systemic Contribution

• Produce the first structured data quality report covering error volumes, trends, and recommended interventions
• Contribute at least one improvement to the validation process: a new validation rule, a refined review checklist, or a structured escalation protocol for a specific error type
• Demonstrate that your corrections are making a measurable difference: model accuracy in your reviewed categories improving, or recurring issues that you surfaced being resolved in the platform

Required Qualifications

• Experience in data analysis, ideally with exposure to financial data
• Strong understanding of financial statements: profit and loss, cash flow statement, balance sheet, and the relationships between them
• Comfortable working with structured data in spreadsheets and SQL: filtering, aggregating, cross-referencing, and spotting anomalies in large data sets
• High attention to detail: able to identify a misclassified account entry, a wrong sign in a cash flow calculation, or an AI-extracted value that does not match the source document
• Analytical rigour: able to distinguish between a data error and a business reality, and to communicate findings precisely without overstating or understating the issue
• Clear written communication: comfortable documenting findings, writing error reports, and providing structured feedback that engineers and AI teams can act on

Nice to Have

• Experience working with accounting or ERP systems (Xero, QuickBooks, Zoho Books, Odoo, Sage) — either as a user, analyst, or integrator
• Familiarity with AI or ML output review: understanding of what hallucinations, confidence scores, and extraction errors look like in practice
• Experience in credit analysis, lending, or financial due diligence — particularly with SME financials
• CPA, ACCA, or equivalent accounting qualification, or active pursuit of one
• Exposure to data quality frameworks, validation rule design, or structured QA processes
• Ability to write basic SQL queries for data investigation and anomaly detection
• Experience with data visualisation tools (Metabase, Tableau, Google Data Studio, or similar) for building operational dashboards

Our Working Style

• Detail-first: in financial systems, a small error in a source metric can cascade into a wrong credit decision — accuracy is the job, not a nice-to-have
• Feedback-driven: your corrections and observations are not just operational — they make the platform smarter; we treat every structured piece of feedback as an engineering input
• Collaborative across disciplines: you will work with engineers, AI developers, and operations staff — translating between data realities and technical solutions is a core part of the role
• Continuous improvement: we expect you to notice patterns, propose fixes, and help build the validation infrastructure that reduces manual review over time
• Fast-paced but careful: we move quickly as a team, but we do not trade accuracy for speed in financial data — if something is wrong, it gets flagged, not waved through

Compensation & Benefits

• KES 30,000- KES 40,000 per month depending on experience
• Clear growth path into senior data, operations, risk, or product roles as Flowt scales
• Direct exposure to AI systems, financial data infrastructure, and the inner workings of a credit platform.
• Hybrid work flexibility and a mission-driven environment