全职 Founding Machine Learning Engineer (LLM) 工作, 薪水, Medicoder 公司招聘中 - Ricebowl

Company Description

Medicoder is a healthcare technology company focused on integrating predictable, reliable AI solutions into clinical and administrative workflows. The organization’s mission is to reduce friction in everyday healthcare work, tackling administrative burnout and elevating operational efficiency across health systems. Medicoder builds intelligent systems that apply advanced AI to large volumes of complex medical documentation, converting raw clinical records into structured, audit-ready insights in seconds. Its proprietary, privacy-first technology helps healthcare organizations securely mask sensitive patient data, streamline high-pressure workflows, and enhance the productivity of care and administrative teams worldwide.

Role Description

As a Machine Learning Engineer specializing in Large Language Models (LLMs) at Medicoder, you will operate at the intersection of state-of-the-art AI engineering and critical, real-world deployment. Every single model you select, benchmark, fine-tune, or train from scratch serves a singular purpose: to drive the success of our core product and bring tangible, measurable value to our customers.

You will own the end-to-end lifecycle of LLMs and deploy optimized, privacy-first solutions directly into production within local hospitals. You will work with a diverse data ecosystem, including public benchmarks, proprietary datasets, and highly sensitive clinical data. Because our systems interface directly with critical hospital infrastructure, you will balance maximizing raw model capability with the strict requirements of data privacy, low-latency performance, and product reliability.

Evidence of Exceptional Ability

At Medicoder, we value raw capability, problem-solving velocity, and execution over mere years of industry experience. In your application, please provide clear evidence of exceptional ability. This does not need to be purely academic or technical. It could be a highly optimized open-source contribution or a breakdown of a complex model you trained from scratch, but it could just as easily be an unconventional, high-stakes risk you decided to take, a bold decision you made against the status quo, or even an extraordinary failure that yielded invaluable, first-principles insights. We are looking for individuals who don't just follow blueprints, but who possess the conviction to act, experiment, and deliver. Show us how you think, how you execute, and what you can build.

Key Responsibilities

Every capability you leverage and every task you execute is directly tied to advancing our product roadmap and elevating the customer experience:

• Collaborate closely with the product team to deeply understand user requirements, customer pain points, and product vision, translating them into robust, actionable technical specifications for model development.
• Rigorously evaluate and benchmark State-of-the-Art (SOTA) LLMs (both proprietary closed-source and open-source models) against complex, domain-specific medical and administrative tasks to ensure maximum product reliability.
• Fine-tune existing architectures and train domain-specific models from scratch using public, proprietary, and unstructured clinical datasets to directly improve product features.
• Translate AI capabilities into user value by deploying, monitoring, and maintaining high-performance LLM pipelines directly within local hospital environments and clinical workflows.
• Optimize models for real-world constraints, focusing on reducing latency, managing GPU memory footprint, and lowering inference costs so our customers experience a seamless, lightning-fast product.
• Curate, synthesize, and clean high-quality pre-training and fine-tuning datasets from complex, multi-modal medical records to continuously fuel product iterations.

Required Qualifications & Skills

• Proficiency in Python and deep learning frameworks (PyTorch or TensorFlow).
• Direct, hands-on experience with LLM training, fine-tuning methodologies.
• Strong understanding of MLOps pipelines, containerization (Docker), and deploying models to cloud or on-premise GPU environments.
• A product-focused, first-principles thinker who thrives in a fast-paced environment and is deeply motivated by solving systemic healthcare challenges rather than chasing abstract research goals.

Nice-to-Haves

• Prior experience dealing with healthcare data standards (FHIR, HL7) or medical ontologies (SNOMED-CT, ICD-10).
• Familiarity with local inference optimization tools (TensorRT-LLM, vLLM, Ollama).
• Experience taking a raw AI prototype and scaling it to a commercially viable SaaS or enterprise product.

Work Environment & Culture

• As a founding member of the company, you will have a direct hand in defining our machine learning stack, infrastructure standards, and architectural best practices.
• We function in a fast-paced startup environment where you are trusted with critical systems and given the space to steer your technical implementation.
• We operate on deliverable-based working hours, not a rigid 9-to-6 schedule. If you deliver world-class results, we don't count the hours.
• You will collaborate closely with engineering teams, company leadership, and end-clients to bridge the gap between technical possibility and real user needs.
• Enjoy the fulfillment of seeing your development and work rapidly deployed into production, actively assisting clinical teams in real-time.

Why Join Us

• We believe our early builders should share deeply in the value they create. Compensation packages will feature a highly competitive combination of salary, stock options, or a customized mix of both.
• Work with production-grade LLM applications tackling actual healthcare bottlenecks, completely bypassing artificial or "toy" datasets.
• We value execution over hierarchy & bureaucracy. High ownership, direct lines of communication, and an environment optimized for builder velocity.
• Gain deep, specialized experience at the cutting edge of applied clinical AI, LLM optimization, and enterprise deployments in highly regulated environments.