[Remote] Agentic AI Software Engineer
Note: The job is a remote job and is open to candidates in USA. TCP Software is a company that has been providing workforce management solutions for over 30 years. They are seeking an Agentic AI Software Engineer to design and implement AI features, manage inference costs, and ensure safe agent actions while mentoring junior engineers.
Responsibilities
- Design, build, and ship agentic AI features end to end, from detection and decision logic through safe action execution, in a cross-functional agile environment
- Ground agent behavior in deterministic rules engines and governed data so every agent explains its reasoning and never acts outside permitted boundaries
- Build and operate the LLM layer, including prompt architecture, structured outputs, confidence scoring, timeouts, and rules-only fallback paths
- Implement safe patterns for agent actions, including idempotent operations, undo windows, retries with backoff, and immutable audit logging
- Develop statistical detection and baseline logic that turns workforce data into ranked, explainable findings with business impact attached
- Manage inference cost, latency, and model selection from development through production
- Partner with QA on evaluation suites, shadow-mode rollouts, and the trust metrics that gate every agent capability, and mentor junior engineers along the way
- Operate independently, setting priorities and goals while continuing to learn and grow technically
Skills
- 6+ years of professional software development experience, including shipping LLM-backed or agentic features to production
- Deep production experience with C#/.NET, SQL Server, and API design in systems where correctness has financial consequences, such as payroll, billing, or banking
- Experience integrating LLM APIs into enterprise applications, with structured outputs, retrieval grounding, and evaluation-driven iteration
- Working fluency with statistical techniques on time-series data, such as rolling baselines, outlier detection, and threshold tuning
- The ability to understand requirements and solve technical issues without supervision, serving as a technical lead
- Hands-on daily use of AI coding assistants or agentic tools, such as Claude Code, GitHub Copilot, or Cursor
- Production experience with LLM APIs, MCP servers, or internal AI tooling
- An understanding of human-in-the-loop agent design and write-capable AI
- A clear understanding of LLM failure modes, including hallucination, prompt injection, and cost overrun, plus the limits and risks of AI-generated code
- Experience with multi-agent orchestration
- Evaluation harness design
- Statistical anomaly detection
Benefits
- 20 Days of PTO (Paid Time Off)
- 13 days of companywide holidays
- 8 hours to volunteer and impact the community
- Comprehensive benefits (Health/Dental/Vision/ 401K)
- Employee Choice Pre-Tax Benefit
Company Overview