Member of Technical Staff, AI (Systems) Platform

Beacon Holdings•San Francisco, CA

About The Position

Beacon Holdings is building an agentic operating system for the vertical SaaS companies it acquires and operates. The platform addresses operational, value-creation, and deal-sourcing work that is bottlenecked by human attention, using real business data and supporting multi-tenant requirements. Members of Technical Staff build and own platform components end-to-end, including design, implementation, operations, and long-term technical direction. The role is described as equivalent to a Staff Engineer, with a flat organizational structure. Beacon is hiring across two specialties: product and platform engineering, focused on agentic products and their underlying services, APIs, data access, and deployment infrastructure; and AI infrastructure, focused on model serving, retrieval, evaluations, observability, fine-tuning, and deployment into secure and regulated environments. Both specialties work directly with users and ship systems into production. The role may cover agentic products, AI infrastructure, and evaluations and observability. Work includes building AI-native tools for portfolio-company operators; serving and operating models across cloud, VPC, and isolated environments; fine-tuning open-weight models; building retrieval and governed data access; selecting models based on cost, latency, and quality; and creating evaluation, tracing, telemetry, replay, and regression systems. The team emphasizes rapid delivery to real businesses, evidence-based quality, and using agents as part of the engineering workflow.

Requirements

  • At least 5 years of experience building and operating distributed systems relied on by real businesses.
  • Experience with systems such as databases, data platforms, query engines, control planes, or large-scale services.
  • Demonstrated performance-engineering experience, including profiling systems to achieve substantial improvements.
  • Understanding of partial failure, capacity limits, multi-tenant isolation, and schema and API evolution.
  • Ability to write clean code in at least one of Python, Java, Go, C++, Rust, or TypeScript.
  • Experience shipping LLM or agentic systems that people rely on, such as RAG over enterprise data, agentic workflows, or LLM-assisted diagnosis or operations.
  • Regular use of coding agents in the development workflow.
  • Understanding of common LLM-system failure modes, including hallucinated tool calls, quality drift, cost increases, and poorly designed evaluations.
  • Experience working directly with customers, handling escalations, and translating customer feedback into product improvements.
  • Experience as a founding engineer or taking a system from 0-to-1 through general availability.
  • Production operations experience, including carrying an on-call pager, incident command, postmortems, and implementing preventative fixes.
  • Comfort working with ambiguity and helping define the platform's scope.
  • For the product and platform specialty: strength in backend services, APIs and client architecture, internal tools, and deployment infrastructure, with enough frontend range to ship a full product; experience building products where the agent is central to the experience.
  • For the AI infrastructure specialty: hands-on experience with model serving such as vLLM or llama.cpp, retrieval and vector search, evaluations, fine-tuning, and model routing.
  • For the AI infrastructure specialty: experience deploying into environments not fully controlled by the team, such as customer VPCs or regulated and isolated regions.
  • For the AI infrastructure specialty: ability to reason about GPU memory, batching, and quantization as well as API design.

Nice To Haves

  • Experience building on or inside modern data platforms such as Iceberg, lakehouse architectures, Snowflake, Databricks, data catalogs, or governance systems.
  • Performance engineering at scale, including profiling, capacity planning, and measurable cost reductions.
  • Experience shipping into regulated environments, including FedRAMP, government or sovereign regions, or air-gapped deployments.
  • Experience building LLM-assisted observability or root-cause analysis over real production telemetry.
  • Experience operating open-weight inference in production, including batching, KV cache, quantization, speculative decoding, or multi-GPU serving.
  • Founding engineer, co-founder, or first-hire experience.
  • Prior Staff or Principal Engineer experience at a high-bar engineering organization.
  • Open-source contributions to data infrastructure, inference, evaluations, or developer tooling.

Responsibilities

  • Own a platform area end-to-end, including its design, implementation, operations, and long-term technical direction.
  • Take user problems from initial conversations through implementation and production deployment.
  • Build agentic products for portfolio-company operators, including agent loops, user interfaces, internal APIs, and deployment infrastructure.
  • Build and operate model-serving systems, including open-weight models, across cloud, VPC, and isolated environments.
  • Fine-tune open-weight models on proprietary data when general-purpose models are insufficient.
  • Build retrieval and governed data-access systems that ground agents in real business data.
  • Select and route models based on explicit trade-offs among cost, latency, and quality.
  • Build evaluation and observability systems, including workflow-level evaluations, traces, telemetry, replay, and regression suites.
  • Use evaluation and observability signals to identify failures and accelerate root-cause analysis.
  • Measure system quality with data and ensure model or prompt changes are evaluated before release.
  • Build products that users adopt and rely on, rather than focusing on demos or launches.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service