The Role
Build the Context Management API and semantic modeling infrastructure for AI systems: design retrieval and data-access patterns, orchestrate LLMs and multi-agent systems, implement self-healing data models and evolving knowledge graphs, and integrate enterprise data sources for reliable AI-driven GTM execution.
Summary Generated by Built In
Founding Full-Stack Engineer
ABOUT DEEPLINE
THE PROBLEM
Location: New York City
Type: Full-time
Deepline is the operating system for GTM execution. We turn operator intent into governed execution and measurable outcomes. Not more dashboards. Not more automations. The backend for GTM engineering that makes your GTM stack actually work. Our vision is "ambient automation" that exists & solves problems before you know they exist.
We're building the universal API for B2B businesses.
Replace 20+ API calls to dozens of tools with a single call to Deepline's Context API.
Deepline is a context manager that understands how the real-world works, with an intent compiler that turns context & natural language into outcomes with guardrails and observability.
- Team: Small senior team from Uber, Lyft, OM1, Capchase. MIT, Waterloo, Berkeley, Princeton, UCSD.
- Funding: $3.3M pre-seed from Lerer Hippeau, K5 Global, Exceptional Capital, Sabrina Hahn, Rohan Shah
Every AI tool today hits the same wall: they can't reliably purchase access to proprietary data, your company's knowledge base, or what they don't already know how to get. Claude Code can't query your Snowflake out-of-the-box. ChatGPT doesn't know your weird custom Salesforce schema. They hallucinate because they lack structured context.
The root cause: data infrastructure was built for humans, not AI agents reasoning about business context. Database access patterns are shifting. SQL won't be how AI systems query data in five years. 3
You'll build the structured context management layer that makes AI context selection reliable in production. This isn't better RAG or fine-tuning. This is inventing new data access patterns and context architectures that power the next generation of AI applications in the fastest changing space around.
WHAT YOU'LL BUILD
Context Management API
Build the context layer AI systems need. Systems that maintain structured context across workflows, self-heal when data changes, and compound knowledge over time.
New Data Access Patterns
Design semantic query interfaces that replace SQL for AI agents. Build retrieval pipelines that reason about context before querying. Create systems that understand business semantics and go beyond data schemas.
Self-Healing Data Models
Architect feedback loops that automatically improve data models based on usage. Systems that detect when context breaks and fix it automatically. Knowledge graphs that evolve as the business evolves.
Semantic Modeling Infrastructure
Users need to be able to improve/expand their data model without data experts. Build the semantic layer that translates business questions & existing reports into precise, verifiable queries. Identity resolution across 50+ enterprise systems. Systems that learn customer language patterns and map them to business outcomes.
WHAT WE'RE LOOKING FOR
Required
• 3+ years building production systems
• Experience with retrieval systems, embeddings, vector databases, LLMs or knowledge graphs
• Production ML experience: monitoring, versioning, evaluation frameworks
• Experience with LLM orchestration (LangChain, LlamaIndex) and multi-agent systems
• Familiarity with semantic layers (dbt), data warehouses (Snowflake, BigQuery), enterprise data systems
Nice to Have
• Enterprise data systems (Snowflake, BigQuery, Salesforce, Segment, Gong)
• Multi-agent systems (LangGraph, CrewAI) or workflow orchestration (Airflow, Prefect)
• Knowledge graphs, graph databases, semantic layer tools (dbt, Cube)
• Real-time data pipelines and streaming architectures
TECH STACK
Core: Python (primary), TypeScript/JavaScript, SQL
LLMs: Anthropic Claude API, OpenAI, in-house frameworks
Knowledge Graphs/RAG
Data Infrastructure: Snowflake/BigQuery/Redshift, dbt, Kafka/Pulsar, Reverse ETL (Hightouch, Census)
Enterprise Integrations: Salesforce, HubSpot, Segment, Gong, Slack, Zendesk, Mixpanel/Amplitude
COMPANY CONTEXT
Stage: $3.3M pre-seed, proven product-market fit, growing adoption
Team: Small senior team from Uber, Lyft, OM1, Capchase. MIT, Princeton, UCSD. You'll be engineer #5-6. Direct collaboration with founders & customers
Culture: First-principles debate. Ship multiple times a day. Rapid iteration. In-person in NYC with quarterly off-sites.
Compensation: $140K-220K base + meaningful equity. Early-stage upside in proven company.
Jai, Saf, & Chirag
Co-founders of Deepline
CompensationThe base pay range for this role is $140,000 – $260,000 per year.
Skills Required
- 3+ years building production systems
- Experience with retrieval systems, embeddings, vector databases, LLMs or knowledge graphs
- Production ML experience: monitoring, versioning, evaluation frameworks
- Experience with LLM orchestration (LangChain, LlamaIndex) and multi-agent systems
- Familiarity with semantic layers (dbt), data warehouses (Snowflake, BigQuery), enterprise data systems
- Core tech stack: Python, TypeScript/JavaScript, SQL
- Enterprise data systems (Snowflake, BigQuery, Salesforce, Segment, Gong)
- Multi-agent systems (LangGraph, CrewAI) or workflow orchestration (Airflow, Prefect)
- Knowledge graphs, graph databases, semantic layer tools (dbt, Cube)
- Real-time data pipelines and streaming architectures (Kafka, Pulsar)
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
Deepline is a GTM (Go-to-Market) data infrastructure platform that provides a unified CLI and API for waterfall enrichment, lead discovery, signal detection, and campaign activation. It helps revenue teams automate their go-to-market stack by integrating with tools like Claude Code, Cursor, and Codex, allowing users to turn data into actionable insights without needing new technical resources.
