Note: The job is a remote job and is open to candidates in USA. Agentic Dream is seeking a Technical Business Analyst to bridge the gap between business intent and actionable specifications for their data projects. The role involves deconstructing complex business processes into clear requirements, collaborating with various stakeholders, and ensuring that data solutions are built correctly. Responsibilities include authoring specifications, maintaining data dictionaries, and running requirements workshops.
Responsibilities
- Deconstruct complex business processes and user stories into EARS-formatted requirements (Easy Approach to Requirements Syntax) that are unambiguous, testable, and traceable
- Author Gherkin scenarios (Given/When/Then) that define the "Definition of Done" — including for non-deterministic agent behavior, where you specify the logic path, not just the output
- Maintain the Master Contextual Package (MCP) traceability: every requirement maps to a business need and to the agent/pipeline logic that implements it
- Build and maintain data dictionaries, schema specifications, and source-to-target mappings for the Bronze → Silver → Gold pipeline and the PostgreSQL schema registry
- Define field-level validation rules, tolerances, and alias dictionaries (e.g., "Qty Shipped" → quantity) that drive the Schema Matching and validation MCP servers
- Profile source data, quantify data-quality issues, and write the acceptance criteria for data-quality gates and cross-field validation (e.g., line items sum to total)
- Translate business policy into structured, machine-consumable business-rule definitions (discount limits, tax rules, workflow steps) and the guardrail specs that enforce them ("Block discount > 20%")
- Maintain the knowledge-pack catalogs (business rules, guardrails, knowledge base) as living, versioned specifications
- Map current-state and future-state quote-to-install and document-processing workflows, identifying automation opportunities, HITL (Human-in-the-Loop) decision points, and exception paths
- Define confidence-based routing rules with stakeholders (≥90% auto-approve, 60–89% review, <60% Expert-in-the-Loop) and the SLAs around them
- Run requirements workshops and translate technical complexity (why an automation is risky, costly, or non-deterministic) into language non-technical stakeholders can act on
- Own UAT planning and acceptance: build test cases from the Gherkin specs, coordinate sign-off, and verify delivered work matches the spec
- Produce operational reporting specs — defining the metrics (throughput, accuracy, cost-per-document, net savings) and dashboards the business needs
- Run discovery sessions with business stakeholders and engineers; produce EARS requirements, Gherkin acceptance criteria, and data mappings
- Write SQL queries to profile and validate data, quantify edge cases, and back requirements with evidence (not assumptions)
- Maintain the requirements backlog and traceability matrix in Jira (project SDB) and Notion
- Partner with Data Engineers on schema/registry specs and source-to-target mappings
- Partner with AI Engineers to define agent behavior, tool contracts, and the "Definition of Done" for probabilistic outputs
- Work with the PM team to scope sprints, estimate, and sequence delivery
- Build and execute UAT test cases from the specs; document defects with clear reproduction steps
- Validate that delivered pipelines and agents meet acceptance criteria before release
- Maintain decision logs and ADR-style records of requirement changes
Skills
- 5+ years as a Business Analyst / Technical BA, with 3+ years on data-intensive projects (data platforms, integrations, analytics, or document/IDP processing)
- Strong SQL — able to independently profile data, validate hypotheses, and write moderately complex queries against PostgreSQL or similar
- Demonstrated requirements engineering skill — user-story decomposition, acceptance criteria, and at least one formal technique (EARS, BDD/Gherkin, use cases, or equivalent)
- Data modeling literacy — able to read and contribute to data dictionaries, schema specs, ER diagrams, and source-to-target mappings. Understands normalized vs. denormalized models conceptually
- Process modeling — BPMN or equivalent; mapping current/future-state workflows and identifying automation and exception paths
- Familiarity with the modern data & AI stack — understands what RAG, vector databases, APIs/microservices, and LLM-based automation are and where they fit (you don't have to build them, but you must spec for them)
- Tooling — Jira, Confluence/Notion, and diagramming tools; comfortable maintaining traceability matrices and living documentation
- English proficiency: B2+ required (C1 preferred). You'll facilitate workshops, write specs, and translate between business and technical audiences daily
- Experience specifying agentic / LLM-driven systems — defining 'Definition of Done' for non-deterministic outputs and writing guardrail/policy requirements
- Hands-on with document/IDP projects — POs, ACKs, Invoices, schema/alias matching, confidence scoring, and HITL review workflows
- Familiarity with Spec-Driven Development (SDD) and the Model Context Protocol (MCP) concept
- Experience writing YAML/JSON business-rule definitions or configuration-as-spec
- Light scripting (Python) for data profiling and analysis
- BI/reporting spec experience (QuickSight, Power BI, Metabase)
- Background in commercial furniture, logistics, distribution, or manufacturing operations
- Domain exposure to quote-to-install / order management processes
Company Overview
We help enterprises become AI-native by embedding the operating model, culture, teams, and processes required to design, deliver, and scale AI-powered systems across the business. It was founded in 2013, and is headquartered in Fort Lauderdale, Florida, US, with a workforce of 51-200 employees. Its website is https://www.agenticdream.com/.