About Tax OS

Tax OS is the autonomous global tax compliance platform. The platform combines a microservices core, an AI-native classification and interpretation layer, and a jurisdiction-aware calculation and filing engine.

You will work directly under the leadership and guidance of the founders driving product and technical direction. This is a small, trust-based team. You will have significant ownership, direct access to the founder, and a real say in how the platform is built.

The Role

We are looking for one senior software engineer who can take a well-defined product specification and turn it into a running, production-grade system. You will be a founding builder: writing the core of the platform, making architectural decisions with the rest of the founders, and shipping the MVP through to the first real filings. This is a hands-on role. You will spend most of your time writing code, reviewing your own work rigorously, and owning features end-to-end — from data model to deployed service.

We are not looking for someone to manage a team. We are looking for someone who builds.

What You Will Build

  • The core microservices of the Tax OS platform: Legislation service, Calculation Engine, User service, Integration service, and API Gateway.
  • The AI-assisted interpretation layer that parses tax legislation and converts it into computable rules, with a human-in-the-loop review workflow.
  • Jurisdiction-aware calculation logic, starting with the Phase 1 jurisdictions and extending into multi-state and multi-country coverage.
  • Integrations with accounting systems, bank feeds, and government filing gateways.
  • The web application and API surface that clients — accountants, SMEs, and firm partners — will actually use.
  • The infrastructure, CI/CD pipeline, observability, and security controls that make the system safe to run with real taxpayer data.

Technical Environment

The reference stack for Tax OS is:

  • Backend: Python 3.12 (FastAPI) for calculation, legislation, and interpretation services; Node.js 20 (NestJS) for user, integration, and notification services.
  • Frontend: React 18 / Next.js 14 with TypeScript; React Native for mobile.
  • AI/ML: LangChain, LLM orchestration, retrieval and semantic search over legislation.
  • Data: PostgreSQL 16, MongoDB 7, Redis, Elasticsearch.
  • Infrastructure: AWS, Kubernetes (EKS), Kafka for events, OpenTelemetry + Prometheus/Grafana for observability.
  • Security: OAuth 2.0 / OIDC, encryption at rest and in transit, audit logging, SOC 2 readiness.

We are pragmatic. If you have a strong reason to adjust a choice for the first 12 months, we want to hear it. What we care about is that the system is correct, auditable, secure, and fast enough to meet the SLAs in the PRD.

Who You Are

  • 3-5 years of professional software engineering experience, with a clear track record of shipping production systems that handle sensitive data or regulated workloads (fintech, tax, healthcare, payments, legal tech, or similar).
  • Strong in at least one of Python or TypeScript/Node, and comfortable moving between both. Comfortable owning backend services, data models, and the APIs that sit on top of them.
  • Comfortable with cloud-native architecture: Kubernetes, event-driven services, and the operational discipline that comes with running them.
  • Practical experience integrating or building with modern AI/ML tooling — LLM APIs, embeddings, retrieval pipelines, or classification models. You do not need to be a researcher; you need to ship AI features that behave well in production.
  • You write clean code, you test what matters, and you care about correctness. For a tax platform, a silent 0.5% calculation error is a serious failure.
  • You can work independently. Remote-first work over a 12+ month horizon requires someone who sets their own pace, communicates clearly in writing, and does not need to be managed.

Trust and Alignment

This is the most important section of the description, and we want to be direct about it.

Tax OS is being built by a small team under a founder. The intellectual property — the architecture, the models, the legislation corpus, the calculation rules — is the company. We need someone who understands that, takes it seriously, and moves in the same direction as the rest of the team.

In practice, that means:

  • You will sign a mutual non-disclosure agreement covering all product, business, and technical information before any confidential material is shared. Confidentiality obligations survive the engagement.
  • You will sign an intellectual property and inventions assignment agreement. All code, models, designs, documentation, and derivative works created in connection with the role are the sole property of the company.
  • You will agree to reasonable non-solicitation and non-competition terms scoped to the tax compliance and tax automation space, for a defined period after the engagement ends.
  • You will not take, retain, or reuse company code, data, models, or documents outside the engagement. On separation, all company material is returned or destroyed and access is revoked.
  • You will treat disagreements as something to work through with the fractional CTO directly, not as a reason to fork direction quietly. We want a builder who flags concerns early and commits once a decision is made.

We are protective of what we are building because we believe it is genuinely valuable. The right person will read this section and recognize themselves in it.

Engagement and Compensation

  • Location: Fully remote within Europe or the United States. You choose your working hours within reason, provided there is meaningful overlap with the team for real-time collaboration.
  • Schedule: Flexible, but this is a full-time role. We expect your primary professional focus for at least 12 months and beyond.
  • Base salary: Competitive cash compensation, benchmarked to senior engineering roles in your region. Paid on a regular cadence.
  • Equity: Shares in the company, granted and vesting based on progress against defined milestones (Phase 1 delivery, first filings, Phase 2 readiness). The equity grant is meaningful — this is a founding-builder role, not a contractor role.
  • Duration: 12-month minimum commitment. The expectation on both sides is that this becomes a long-term role if the first year goes well.

How to Apply

Send a short note directly to us and include:

  • A brief description of one or two production systems you have personally built that are comparable in complexity — what the system did, what you owned, and what you would do differently today.
  • Links to code, technical writing, or projects that represent your work. A working prototype or sample matters more than a polished résumé.
  • Your location, notice period, and target base salary range.
  • One sentence on why Tax OS in particular — not why startups in general.

We will reply to every serious application. First conversations are with the fractional CTO directly.

Build AI That Makes Accounting Autonomous

We’re creating AI that handles the entire complexity of accounting workflows autonomously – from transaction processing to report preparation, from reconciliation to compliance checks. The accountant remains for final review and filing, but the heavy lifting becomes entirely autonomous.

Our approach: continuous learning from millions of real transactions creates compound intelligence that no static model can match. Every transaction teaches our system. Every correction improves tomorrow’s decisions. Every client pattern becomes permanent knowledge. While others retrain models quarterly, our AI evolves every second from actual usage.

What You’ll Build

  • Continuous Learning Architecture: Systems that turn millions of daily transactions into compound intelligence without model retraining.
  • Autonomous Workflow Systems: AI that completes complex, multi-step processes independently – categorization, reconciliation, report generation.
  • Client Intelligence Layers: Architecture where each business’s AI becomes uniquely accurate through learned patterns.
  • Industry-Defining Benchmarks: Evaluation frameworks that establish what autonomous accounting means.

You’ll work directly with leading AI labs – Anthropic, OpenAI, or others – to build custom capabilities that don’t exist in any current API.

Why This Matters

We’re eliminating the drudgery that consumes millions of hours – the manual categorization, the endless reconciliation, the repetitive compliance checks. Accountants should focus on strategy and decisions, not data entry and matching. When this works, accounting becomes a strategic function, not an operational burden.

The Challenge

Building AI that handles real money autonomously requires solving unsolved problems:

  • Online learning with financial-grade accuracy guarantees
  • Client-specific optimization at million-user scale
  • Continuous improvement without catastrophic forgetting
  • Explainable decisions for audit trails and compliance

These aren’t research questions – you’ll solve them in production, processing billions in real transactions.

What You Bring

  • Track record building AI systems that redefined what’s possible.
  • Deep expertise in online learning, continuous adaptation, and memory architectures.
  • Experience shipping complex systems at scale with real-world constraints.
  • Ability to partner with AI labs to push beyond current capabilities.
  • Vision for how AI achieves true autonomy in professional work.

The Opportunity

Define the architecture that makes professional work autonomous. Build systems that transform accounting from manual labor to strategic oversight. Create the intelligence layer that makes continuous learning possible at scale.

You won’t implement our vision – you’ll create it.

4 min. read Posted on 07/02/2024

Connect Every Financial System to Fiskl

We’re at 80% automation today, building toward 95% autonomous accounting. We need someone obsessed with making every transaction – from every payment provider, bank, and platform – flow seamlessly into Fiskl. Zero manual entry is the goal.

The Mission

Build bulletproof integrations that capture every transaction from every source. We’ve nailed Stripe – now we need everything else. When a business uses Fiskl, every penny should appear instantly, ready for our AI to reconcile.

What You’ll Build

  • Payment Integrations: PayPal, Square, Adyen, Mollie, GoCardless, Wise, crypto wallets.
  • Banking Feeds: Open Banking APIs, Plaid, TrueLayer – real-time transaction streaming.
  • E-commerce Platforms: Shopify, WooCommerce, Amazon Seller, eBay.
  • Accounting Migrations: One-click migration from Xero/QuickBooks.
  • Data Enrichment: Transform messy transaction data for 98%+ AI matching accuracy.

Non-Negotiable: You Must Code with AI

We only hire engineers who use AI tools to ship faster. You should be using Claude/GPT-4 to parse API docs, generate boilerplate, and handle edge cases. If you code without AI, don’t apply.

Technical Requirements

  • 3+ years building production payment/banking integrations.
  • Strong backend development skills (Java preferred, but we care more about integration expertise).
  • Deep understanding of payment flows, settlement, and reconciliation.
  • Experience with Open Banking/PSD2 standards.
  • Event-driven architecture and distributed systems experience.
  • Track record of building resilient, self-healing integrations.

Technical Environment

  • Java backend services with AWS infrastructure.
  • Event processing with EventBridge and Step Functions.
  • Real-time streaming with Kinesis and SQS.
  • Lambda for serverless webhook processing.
  • Integration with AWS Bedrock for AI-powered reconciliation.

The Reality of This Role

  • Parse wildly inconsistent data across 100+ providers.
  • Handle multi-currency complexity and settlement timing issues.
  • Build for 99.99% reliability when third-party APIs fail constantly.
  • Reverse-engineer undocumented APIs when needed.
  • Obsess over edge cases that affect 0.01% of transactions.

What Makes You Successful

  • You’re obsessed with data completeness – 99.9% isn’t good enough.
  • You understand that missing £0.01 is unacceptable in accounting.
  • You build systems that self-heal rather than page you at 2am.
  • You use AI to accelerate everything from API exploration to testing.
  • You think in failure modes and edge cases.

The Opportunity

The accounting industry is a $88B market with data trapped in thousands of silos. Every business uses 5-10 different financial tools that don’t talk to each other. Manual data entry costs billions in hours and errors.

This is your chance to build the connective tissue that makes autonomous accounting possible. Every integration you ship unlocks an entire market segment. A successful Shopify integration means 2 million merchants can automate overnight.

Why This Matters

You’re not just moving data between systems. You’re eliminating the most painful part of running a business. Every integration saves thousands of hours of manual work. Every edge case you handle prevents a business owner from panicking about missing money.

Interview Process

  • Review of production integrations you’ve built
  • Live coding session building an integration with AI assistance
  • Architecture discussion on handling distributed transactions
  • Deep-dive on your approach to reliability and error handling

What We Offer

  • Competitive salary based on experience.
  • Own the entire integration architecture.
  • Process millions in real transactions daily.
  • Work with our AI team to push reconciliation accuracy to 100%.
  • Remote-first with quarterly team sprints.

To Apply

Show us:

  • Payment or banking integrations you’ve built.
  • How you use AI tools to accelerate integration development.
  • Your approach to handling edge cases and failures.
  • The worst API you’ve conquered.

If you think “good enough” integrations are acceptable, don’t apply. If you’re obsessed with capturing every transaction perfectly and making unreliable APIs bulletproof, let’s talk!

Build Interfaces That Make AI Feel Like Magic

We’re at 80% automation today, building toward 95% autonomous accounting. Now we need someone to make that complexity feel simple. If you believe the best interface is no interface, and that AI should anticipate needs not wait for clicks, this is your role.

The Mission

Create React interfaces where AI does the heavy lifting and humans just approve. Build experiences so intuitive that accountants who still use Excel become power users in minutes.

What You’ll Build

  • Real-time dashboards that update as AI processes millions of transactions.
  • Conversational interfaces where SME owners ask “what can I afford?” and get instant answers.
  • Intelligent forms that complete themselves using AI predictions.
  • Mobile-first experiences that work flawlessly on any device.
  • Data visualizations that make complex financial data instantly understandable.
  • Collaborative interfaces where accountants and clients work simultaneously.

Non-Negotiable: You Must Code with AI

We only hire engineers who are highly proficient with AI tools. You should be using Claude/GPT-4/Copilot daily to ship 3-5x faster than traditional development. If you don’t actively code with AI, this role isn’t for you.

Technical Requirements

  • Production React experience (we care more about what you’ve built than years)
  • Strong TypeScript skills and modern React patterns (hooks, context, suspense)
  • Production experience with React frameworks (Next.js, Vite, Remix, etc.)
  • Proven proficiency with shadcn/ui and component composition
  • Experience integrating with backend APIs and real-time data
  • Strong understanding of performance optimization
  • Eye for design and UX without needing hand-holding

Our Stack

  • React + TypeScript for everything frontend
  • shadcn/ui for components, open-source libraries where they excel
  • Vercel for deployment and edge functions
  • AWS Bedrock for AI integrations
  • Real-time updates via WebSockets

What Makes You Different

  • You ship production code with AI assistance daily.
  • You compose existing libraries rather than reinventing wheels.
  • You can turn complex requirements into simple interfaces.
  • You think in user journeys, not just features.
  • You understand Vercel’s deployment model and optimize for it.
  • You balance perfectionism with shipping.

The Opportunity

The accounting industry is a $88B market where the “modern” interface is still a digitized paper ledger. QuickBooks looks like Windows 95. Xero thinks tabs are revolutionary.
This is your chance to build the interface that makes AI-powered accounting feel inevitable, not intimidating. To create experiences so good that switching back to traditional software feels like going back to dial-up internet.

Interview Process

  • Portfolio review focusing on production React applications.
  • Live coding with AI tools to build a financial dashboard component.
  • Discussion of your AI-assisted development workflow.
  • Technical deep-dive on performance and architecture decisions.

What We Offer

  • Competitive salary based on experience.
  • Build the future of AI-powered accounting interfaces.
  • Work with cutting-edge AI models and infrastructure.
  • Remote-first with quarterly team meetups (not quarterly sprints!).
  • Direct impact on millions of global users.

To Apply

Show us:

  • Production React applications you’ve built.
  • How you use AI tools to accelerate development.
  • A shadcn/ui based component or project.
  • Your approach to making complex data simple.

If you want to build another CRUD app with forms and tables, this isn’t for you. If you want to create interfaces that make people rethink what software can do, let’s talk!

Build Autonomous Systems That Actually Work

We’re at 80% automation today, building toward 95% autonomous accounting. Looking for engineers who want to ship AI that eliminates entire categories of human work.

The Mission

Build autonomous agents that complete entire accounting workflows without human intervention. Your code will power self-running reconciliation, intelligent categorization, and eventually full month-end close. This isn’t about incremental improvement – it’s about making manual bookkeeping obsolete.

What You’ll Build

  • Self-improving reconciliation agents that match transactions across 170+ currencies, crypto wallets, and payment processors with 98%+ accuracy.
  • Multi-agent orchestration systems that coordinate complex workflows – from invoice processing to financial reporting.
  • Document intelligence pipelines that extract and process any financial document.
  • Reasoning agents that detect anomalies, explain decisions, and self-correct.
  • Real-time processing systems handling millions of transactions with sub-second response times.

Non-Negotiable: You Must Code with AI

We only hire engineers who are highly proficient with AI tools. You should be using Claude/GPT-4/Copilot daily to ship 3-5x faster than traditional development. If you don’t actively code with AI, this role isn’t for you.

The Technical Challenge

  • Build agents that improve themselves – learning from corrections and patterns.
  • Handle infinite edge cases across 21,000+ bank feeds and payment methods.
  • Create explainable AI that accountants trust with their clients’ finances.
  • Design fault-tolerant systems that self-heal and never lose a transaction.
  • Scale from SMEs to enterprise volumes without architecture changes.

Core Technical Work

  • Implement advanced RAG systems with domain-specific financial knowledge.
  • Build custom model training pipelines using AWS SageMaker.
  • Design event-driven architectures for real-time agent coordination.
  • Create feedback loops where agents learn from each correction.
  • Develop testing frameworks for autonomous systems with real financial implications.

What You Need to Show Us

  • You’ve built real AI/ML systems – production, research, or compelling personal projects.
  • Strong Python skills and experience with modern AI frameworks.
  • Hands-on work with LLMs, agent architectures, or autonomous systems.
  • You ship working code, not just experiments.
  • You can reason through complex system design problems.

Technical Environment

  • AWS AI infrastructure: Bedrock, SageMaker, Claude, Nova models.
  • Distributed processing handling millions of daily transactions.
  • Real-time event streaming and complex orchestration.
  • Multi-region deployment, 24/7 uptime requirements.
  • Python-heavy backend with modern ML infrastructure.

What Makes You Successful Here

  • You’re obsessed with automation – doing anything manually twice bothers you.
  • You think in systems – how agents work together, not just individually.
  • You measure success by customer hours saved and errors prevented.
  • You’re comfortable with ambiguity and can define your own path.
  • You want to prove what AI can actually do beyond demos and prototypes.

The Reality Check

  • This is hard – you’re automating work that humans have done for centuries
  • You’ll deal with real money, real businesses, and zero tolerance for errors
  • The edge cases are endless and sometimes bizarre
  • You’ll need to learn accounting concepts (we’ll teach you)
  • Your code will process billions in transactions

The Opportunity

Accounting is a $88B market where most “AI” is just better search or rules engines. We’re building agents that actually complete the work. You’ll prove that autonomous systems can handle complex, high-stakes tasks that everyone said required human judgment.

This is your chance to build AI that runs businesses, not chatbots that talk about running businesses.

What We Offer

  • Competitive salary based on experience
  • Ship to production in your first week
  • Own critical systems from day one
  • Work with a small team where your contributions immediately matter
  • Remote-first with quarterly team meetups
  • Learning budget for courses and resources

To Apply

Show us:

  • A project where you’ve built agents or complex automation
  • Your approach to making systems truly autonomous
  • Code that demonstrates your thinking (GitHub, notebooks, anything)
  • Why you’re excited about replacing human work with AI

If you want predictable work on marginal improvements, this isn’t for you. If you want to build autonomous systems that handle real-world complexity at scale and define what AI can actually do, let’s talk!