What are the responsibilities and job description for the Generative AI Engineer position at Amtex Systems Inc?
Forward Deployment Engineer/ Gen AI Engineer
Fulltime Opportunity
Hybrid from - Hartford, CT or Round Rock, TX
Must Have:
- AI Tooling: Claude, Cursor, Codex; LLM APIs (Anthropic, OpenAI); prompting, tool use, agent patterns, MCP
- Frontend: React, TypeScript
- Backend: Python or Node.js, REST/GraphQL APIs, event-driven service design
- Data Engineering Databricks: (PySpark, Delta Lake, notebooks, workflows)
- Cloud/Infra: AWS (S3, Lambda, Glue, Redshift), Infrastructure-as-Code (plus)
- BI/Visualization: Streamlit, Tableau, Evidence (nice to have)
Description:
- Build and ship AI agents and automation harnesses as a core deliverable — not a side experiment — using tool use/function calling, multi-turn context management, and agentic design patterns (MCP, LangChain-style frameworks)
- Use Claude, Cursor, and Codex as your primary development environment daily — build with AI, not around it, across every layer you touch
- Evaluate and correct non-deterministic model output as a first-class engineering discipline — know what the AI wrote, where you overrode or discarded it, and what would have shipped broken if trusted blindly
- Take a problem from rough idea to deployed, working software with minimal handoffs — writing code, shaping UX, and wiring data pipelines yourself, accelerated by AI tooling throughout
- Design agent skills and internal AI-assisted workflows that other engineers on the team rely on and build from
- Move across frontend, backend, data engineering, and infra within the same sprint, using AI tools to compress the time each layer normally takes
- Design and maintain data pipelines and analytical surfaces on Databricks and AWS that non-engineers can actually use
- Work directly with product managers and stakeholders — push back on scope, propose better (often AI-driven) solutions, and make pragmatic trade-offs without waiting to be told
- Own architectural decisions for your product area, including when an agent/LLM-based approach is the right call versus deterministic code
- Leave the codebase simpler than you found it — know when to abstract, inline, or simplify rather than add
- Deploy, debug, and operate confidently in AWS without breaking production
- Deliver outcomes that would take a conventional team 5–10x longer — the agentic/AI-native workflow itself is the reason for that multiplier, not just raw coding speed