Senior Founding AI Software Engineer

Deffai; United States; Remote; USD 180,000 - 260,000; Full time

Benefits offered by this job

Equity

Remote within the United States

Path to Head of Engineering

Job summary

Deffai is seeking a founding-level engineer to help build an AI-native platform that simulates regulatory outcomes for therapeutics. The role spans front-to-back, from ingestion and retrieval to agents, evaluations, and infrastructure, with close collaboration with former regulatory reviewers.

Remote within the United States, potential path to Head of Engineering. You will work across React/TypeScript frontends, Python/FastAPI backends, and AWS infrastructure, shaping production-grade systems

Qualifications

• 5+ years building and operating production software and owning production architecture.
• Experience shipping LLM systems to production with document retrieval or agent workflows.
• Daily use of AI coding tools with the ability to audit migrations and data-access code.

Responsibilities

• Build evaluation and benchmark pipelines with regression suites and gating prompts/models.
• Implement ingestion and retrieval over long regulatory documents including parsing and hybrid search.
• Design and ship agent systems, runtimes, and tool integrations for production.
• Strengthen the product surface in React, TypeScript, Vite, Tailwind, and the Vercel AI SDK.
• Own backend and persistence with Python, FastAPI, SQLAlchemy, PostgreSQL, Alembic including idempotent data pipelines.
• Support AWS infrastructure with ECS/Fargate, Terraform, GitHub Actions, and environments with RBAC, encryption, and audit logging.

Skills

Python

Data engineering

LLM production systems

FastAPI

SQLAlchemy

React

TypeScript

AWS

Terraform

GitHub Actions

PostgreSQL

Vercel AI SDK

Tools

PostgreSQL

Alembic

SQLAlchemy

GitHub Actions

Vercel AI SDK

Vite

Tailwind

Terraform

ECS/Fargate

LangGraph

Claude Agent SDK

Job description

Role overview

This is a founding-level engineering role at a venture-backed, AI-native seed-stage startup building an intelligence platform that simulates regulatory outcomes for therapeutics companies and their investors. The work spans the full stack, from document ingestion and retrieval to agents, evaluations, and infrastructure, and is being built in close collaboration with former regulatory reviewers. The role is remote within the U.S., reports into a small engineering team, and carries a viable path to Head of Engineering.

Responsibilities

• Build the evaluation and benchmark pipeline: regression suites, precision and recall measurement against expert judgement, and gating prompt, model, and agent changes
• Build ingestion and retrieval over long, complex regulatory documents (scanned PDFs, Word files, tables, slide decks), including parsing, structure-aware chunking, and hybrid search
• Design and ship agent systems, runtimes, and tool integrations that operate reliably in production
• Strengthen the product surface in React, TypeScript, Vite, Tailwind, and the Vercel AI SDK
• Own backend and persistence work in Python, FastAPI, SQLAlchemy, PostgreSQL, and Alembic, including idempotent and observable data pipelines
• Support AWS infrastructure built on ECS/Fargate, Terraform, GitHub Actions, and separate dev/prod environments, with a focus on tenant isolation, RBAC, encryption, and audit logging

Requirements

• 5+ years building and operating production software, including owning the architecture of a production system
• Hands-on experience building evaluations against expert judgement (golden datasets, precision/recall) and using them to decide what ships
• Experience shipping LLM systems to production, such as retrieval over complex documents or agent workflows
• Daily use of AI coding tools, with the judgement to catch errors in migrations, infrastructure changes, and data-access code
• Strong Python and data engineering skills: schema design, zero-downtime migrations, reliable pipelines, and background jobs on a relational database
• Experience running production systems on a major cloud with infrastructure as code, CI/CD, and separate dev/prod environments

Nice to have

• Working directly with domain experts (clinicians, lawyers, scientists) to turn their judgement into requirements and test cases
• Retrieval techniques such as hybrid search, reranking, parsing scanned PDFs and tables, or knowledge graphs like GraphRAG
• Agent frameworks and runtimes such as LangGraph, MCP, or Claude Agent SDK, including sandboxed agent execution
• Post-training or fine-tuning (SFT, preference tuning, LoRA) and judgement on when it outperforms prompt and retrieval work
• Building secure multi-tenant systems, AWS (ECS, RDS, S3), and Terraform

Benefits and work setup

• Market-standard compensation, equity, and a path toward Head of Engineering or beyond
• Remote within the United States; no visa sponsorship currently offered
• Interview process: 30-minute intro with the CEO/cofounder, a 40-minute technical interview with a short hands-on exercise, a short paid take-home (5-8 hours) with a walkthrough, and a final round with a former regulatory decision-maker; decision targeted within two weeks of final round