AI/ML Engineer (LLM & Agents)

BearPlex; USA; Remote; USD 4.300 - 6.500; Vollzeit

Benefits dieser Stelle

Fully remote work in Pakistan

Learning support for research and tool

Zusammenfassung

BearPlex seeks an experienced LLM Engineer to bring large language model features from design to production. You will build real-world applications, design agents with tool interfaces and memory, and implement robust retrieval and deployment pipelines.

The role emphasizes observability, latency control, cost management, and auditability in regulated environments, with remote work options in Pakistan and opportunities to work with open-weight models and deployment stacks.

Qualifikationen

• Experience shipping LLM features used by real users.
• Strong Python skills and ML tooling familiarity.
• Understanding of agents, tool use, retrieval, context management, and failure modes.
• Evaluation-driven approach to measuring model and system changes.
• Hands-on experience with open-weight models, fine-tuning, and model serving.
• Production discipline around cost, latency, observability, and auditability.

Aufgaben

• Build LLM applications, defining prompts, tool interfaces, and orchestration for real workloads.
• Design agents with tool use, memory, and guardrails that respect operational boundaries and fail safely.
• Develop retrieval pipelines with chunking, embeddings, reranking, and source traceability.
• Create evaluation datasets and offline and online measurements to assess model changes.
• Fine-tune and serve open-weight models, including alignment work when a project requires it.
• Operate production features by managing latency, cost, caching, fallbacks, monitoring, and deployment in client-controlled environments when needed.

Kenntnisse

Python

Agent design

Retrieval

Model serving

Observability

Ausbildung

Bachelor's degree in CS/ML

Tools

Docker

Kubernetes

Node.js

TypeScript

Jobbeschreibung

Role overview

Take large language model features from early design through production, building systems that are measurable, observable, and suitable for use in regulated environments. The role spans agent design, retrieval, model adaptation, and production operations, with attention to traceability, safe behavior, latency, and cost.

Responsibilities

• Build LLM applications, defining prompts, tool interfaces, and orchestration for real workloads.
• Design agents with tool use, memory, and guardrails that respect operational boundaries and fail safely.
• Develop retrieval pipelines with chunking, embeddings, reranking, and source traceability.
• Create evaluation datasets and offline and online measurements to assess model changes.
• Fine-tune and serve open-weight models, including alignment work when a project requires it.
• Operate production features by managing latency, cost, caching, fallbacks, monitoring, and deployment in client-controlled environments when needed.

Requirements

• Experience shipping and maintaining LLM-powered features used by real users.
• Strong, maintainable Python skills and familiarity with ML and data tooling.
• Practical understanding of agents, tool use, retrieval, context management, and common failure modes.
• An evaluation-driven approach to measuring model and system changes.
• Hands-on experience with open-weight models, fine-tuning, and model serving.
• Production engineering discipline around cost, latency, observability, and auditability.

Nice to have

• Experience with preference data, reward modeling, RLHF, or other alignment methods.
• Familiarity with vector stores, PostgreSQL, and embedding pipelines at scale.
• Model deployment experience with AWS or GCP, Docker, and Kubernetes.
• Ability to integrate models into applications using Node.js or TypeScript.

Benefits and work setup

• Fully remote work within Pakistan.
• Learning support for research, courses, compute, and tools.
• Growth opportunities in model engineering and AI architecture.