Fullstack Engineer
Software Engineering
About WithConcord
Our story
After spending 15+ years helping build and run some of the most highly admired companies and venture funds in San Francisco, we have learned some painful lessons that all founders unfortunately experience when building their masterpieces.
We’re the team founders and venture funds call at 11 p.m. on a Friday, when the round has to close Monday and the numbers don’t tie.
We are on a mission to build a company that makes it 10X easier for founders to run their businesses. We have lost sleep over crises, so you don’t need to.
Our vision
We are building a suite of interconnected, AI-native products to help startups with their back-office needs.
Equity is where we start. Over time we want to be the one partner you need across your entire back office, that you can always count on.
About The Role
OpsDNA is building an agentic CFO platform for venture capital firms. We turn the information buried in legal documents and financial records into structured data and workflows that firms can use to run their businesses. A central challenge is document understanding. Partnership agreements contain the facts and rules our platform needs. Those facts are spread across pages, definitions, tables, and cross-references—and getting them right matters. We’re looking for an engineer to own this problem, working directly with the founders to build the systems that take us from a PDF to reliable, usable information. Your primary focus will be building our document classification and extraction capabilities: identifying what a document is, finding the relevant evidence, and extracting structured information for ingestion into OpsDNA. This problem repeats across many use cases. You’ll build reusable capabilities that support different document types and extraction tasks, while handling the details that make each one difficult. You should already be deeply comfortable building software with coding agents. We’re a small team that intends to stay highly productive, and we expect you to explore what becomes possible with capable models, agent orchestration, and isolated execution environments. You’ll have substantial freedom to experiment and clear responsibility for the quality of what ships.
What You’ll Do
•Build document classification and segmentation. Identify document types, separate documents within combined PDF packets, and distinguish operative agreements from attachments, summaries, and supporting material. •Develop reliable extraction pipelines. Combine PDF parsing, OCR, layout information, classifiers, and LLMs to turn document content into validated, typed data. •Find the evidence an extraction task needs. Route relevant clauses to the appropriate extractor, preserving definitions, exceptions, and cross-references needed to interpret them correctly. •Connect every extracted fact to its source. Preserve page and passage references, distinguish missing information from uncertain answers, and support efficient review and correction. •Evaluate and improve system performance. Build realistic evaluation sets and measure classification accuracy, extraction completeness, false positives, cost, and latency. Account for duplicate templates and differences between document sources. •Choose and integrate the right tools. Evaluate services such as Extend and Reducto, models such as Jev, and libraries such as PDFium. Make practical decisions about where to use a vendor, a model, or deterministic code. •Ship the surrounding product. Build the TypeScript services, PostgreSQL data flows, and review interfaces that make these capabilities useful to customers. Help shape how we engineer. Develop effective workflows for coding agents, testing, observability, and code review as part of the founding team.
What You’ll Bring
•Strong software engineering fundamentals. You have shipped substantial software and can take a problem from an unclear requirement through implementation and production debugging. •Deep experience with agentic coding. Tools such as Amp, Codex, Claude Code, Cursor, or comparable environments are part of how you work. You can delegate substantial tasks to agents, manage their context and execution, and evaluate the resulting code. •Hands-on experience with LLM-assisted classification or extraction. You’ve built systems that turn unstructured inputs into useful decisions or structured outputs, and you understand the failure modes that appear beyond a successful demo. •Strong TypeScript and backend skills. You’re comfortable building APIs, data pipelines, and relational data models, with enough full-stack fluency to deliver a complete workflow. •An experimental approach. You can formulate a hypothesis, build an evaluation, inspect failures, and decide whether a change actually improved the system. •Care for correctness and evidence. You understand why schema validation, source provenance, uncertainty, and recovery paths matter when software handles consequential information. •Product judgment and ownership. You work closely with founders and customers, communicate clearly, and make progress through ambiguity. •Prior financial-services experience is not required. Curiosity and a willingness to learn venture capital operations and fund finance are essential. Helpful, but not required •Experience with Jev / TypeSafe or other specialized classification models. •Experience with Extend, Reducto, PDFium, OCR, or document layout analysis. •Experience extracting information from contracts, legal documents, or financial records. •Experience running coding agents in containers or coordinating multiple agents. •Familiarity with Effect, functional programming, PostgreSQL, or multi-tenant applications. How we work Our core stack is TypeScript, PostgreSQL, and Effect. We value explicit contracts, clear failure behavior, and systems whose outputs can be traced back to evidence. You’ll work closely with the founders, have direct context for product decisions, and influence both the technical direction and the way we build. Success means turning difficult documents into dependable product capabilities—and developing an engineering process that lets a small team keep doing it well.
*This job posting exists to fill a vacancy.