Founding AI Engineer (Agent Systems)

Vincer•Boston, MA

1d•Hybrid

About The Position

Vincer is seeking a founding AI engineer to own the intelligence layer of a voice-first AI system for life sciences customers. This role involves shaping the AI direction and future roadmap, working with multi-agent workflows, existing evaluation work, and a working MVP. The company is solving the problem of supporting life sciences sales reps in complex environments where they often make critical decisions without immediate support. Vincer is building a voice-first AI decision support system to provide this support, understanding complex situations, assembling context, retrieving approved knowledge, orchestrating AI systems, and responding conversationally. The product must be reliable, predictable, and safe in a regulated environment.

Requirements

5+ years of full-time software engineering experience.
At least 2 years of recent, hands-on experience building and shipping AI systems in production (not prototypes or demos).
Experience designing or working with agentic systems: routing, tool use, retrieval, context management, and multi-step workflows.
Experience building across multiple model providers.
Located in Boston or NYC.
Willingness to engage in regular, in-person collaboration.

Nice To Haves

Experience in enterprise software delivery, especially in a regulated industry (healthcare, finance, or similar).
Experience being responsible for an AI system after launch, not just through launch.
Comfort encoding expert judgment into measurable system behavior, working alongside domain experts.
Experience with voice or other latency-sensitive interfaces.
Rigor about measurement; preferring smaller, provable improvements over larger, unproven ones.
Understanding of the strengths and limits of model-based evaluation.

Responsibilities

Own Vincer's intelligence layer: agent architecture and orchestration, evaluation, retrieval and grounding, guardrails, and AI quality.
Evolve the orchestration layer for independent agent versioning, deployment, and improvement.
Build evaluation systems to make agent quality visible (failure modes, regressions, improvements).
Build retrieval and grounding over approved customer content, treating unsupported answers as system failures.
Own guardrails, escalation logic, and safety behavior for a regulated environment.
Engineer for latency and cost through context management, caching, model selection, and routing across model tiers.
Encode expert judgment into system behavior, working directly with the scientific co-founder.