Open Positions

AI Developer

Remote

Full-Time

Please include your portfolio and resume.

About the Role

We are seeking a highly skilled and innovative AI Developer to join our team dedicated to building next-generation AI solutions for the scientific research and mental health industry. This role is crucial for bridging the gap between cutting-edge LLM research and robust, scalable production systems. The ideal candidate will possess deep expertise in deploying foundational models, engineering complex retrieval systems, and championing MLOps best practices.

The AI Developer is responsible for the full lifecycle development of our core intelligent systems, specifically focusing on designing, building, and scaling autonomous AI agents and Retrieval-Augmented Generation (RAG) pipelines. This role ensures that our LLMs are accurately grounded in proprietary and scientific knowledge, are optimized for high performance, and are deployed safely and reliably using modern MLOps, containerization, and cloud technologies. The ultimate goal is to transform complex data science prototypes into stable, high-value products that drive our company's core mission.

Key Responsibilities

RAG and Agentic AI

Develop, deploy, and manage autonomous software agents capable of multi-step reasoning, tool-use (function calling), and complex workflow orchestration.
Design and implement end-to-end RAG pipelines for grounding Large Language Models (LLMs) with proprietary and deep research knowledge, ensuring high factual accuracy and minimizing hallucinations.
Develop, implement, and optimize advanced retrieval strategies, including sparse retrieval techniques (e.g., BM25), hybrid search, and hierarchical/Parent-Child chunking to ensure the LLM receives the most accurate, granular context for generation.

Generative AI and Fine-tuning LLMs

Conduct experimentation and fine-tuning (e.g., using LoRA, QLoRA) of open-source LLMs (e.g., Llama, Mistral) for domain-specific tasks, ensuring models align with performance, latency, and memory constraints.
Implement strategies for prompt engineering, chain-of-thought (CoT) prompting, and model output parsing to maximize model utility and reliability.
Implement logic to automatically flag or filter out generated content that contradicts or cannot be grounded in the authoritative corpus, ensuring factual compliance and minimizing harmful inaccuracies.

Core Machine Learning and AI

Transform data science prototypes into scalable systems: Write clean, production-ready code to modularize, optimize, and integrate core machine learning models (NLP, predictive modeling, etc.) into the main application architecture via RESTful APIs (e.g., Flask/FastAPI).
Data Pipeline Development: Design and implement robust data pipelines for model training, feature engineering, and inference, managing large, complex, and potentially unstructured datasets.

MLOps, DevOps, Cloud

Implement and maintain Continuous Integration/Continuous Delivery (CI/CD) pipelines specifically for machine learning models, ensuring reliable and automated testing, versioning, and deployment.
Manage and monitor deployed models (ModelOps), tracking performance metrics, drift detection, and enabling A/B testing of different model versions in a live environment.
Utilize containerization (Docker) and orchestration (Kubernetes) for packaging and scaling model services.
Manage core cloud infrastructure (e.g. AWS, Azure, or GCP) required for training and serving high-availability AI services.

Requirements & Qualifications

Mandatory Skills

Prior experience working with large-scale data, scientific data, or unstructured text.
Expert proficiency in Python and its scientific stack (NumPy, Pandas, Scikit-learn).
Strong command of at least one major deep learning framework, such as PyTorch or TensorFlow, specifically for model training, optimization, and serving.
Hands-on experience with the Hugging Face ecosystem for model loading, fine-tuning (e.g., PEFT/LoRA), and inference.
Proven ability to build scalable and robust RESTful APIs (e.g., using FastAPI or Flask) to serve models for low-latency inference.

Preferred Expertise

Experience in RAG and Vector Stores: vector databases (e.g., Pinecone, ChromaDB, Weaviate), search frameworks (e.g., Elasticsearch/OpenSearch for BM25), and retrieval orchestration libraries (e.g., LangChain or LlamaIndex).
Experience in MLOps and Orchestration: MLflow or DVC for experiment tracking and versioning, and proficiency with a workflow orchestrator like Airflow or Prefect for data and training pipelines.
Experience in Cloud & Infrastructure: leveraging AI cloud solutions (e.g., AWS SageMaker, Azure ML, or GCP Vertex AI) for managed training and serving high-availability models.
Experience in Testing & Validation: techniques for model unit testing, integration testing, and data validation tools (e.g., Great Expectations).

Key Deliverables

Autonomous Agents

Fully deployed, multi-step software agents handling complex workflows.

Production RAG Pipeline

Operational Retrieval-Augmented Generation (RAG) system demonstrating high factual accuracy, using advanced retrieval like BM25 hybrid search and Parent-Child chunking.

Optimized LLMs

Domain-specific, fine-tuned LLMs that meet production latency and memory constraints.

Factual Guardrails

Automated logic and systems implemented to flag, filter, or correct generated content that contradicts the authoritative knowledge base, ensuring factual compliance.

Production Code

Modular, clean, and optimized code that successfully transforms data science prototypes into reliable, scalable production systems.

Data Pipelines

Robust and automated data pipelines for model training, feature engineering, and inference management.

Automated CI/CD

Functional Continuous Integration/Continuous Delivery (CI/CD) pipelines for automated model testing, versioning, and deployment.

ModelOps Monitoring

Implemented ModelOps systems for tracking performance, drift detection, and enabling A/B testing in production.

Workplace Policies

confidential Information

Definition:

Technical documentation, source code, AI models, and project specifications
Contracts, agreements, and vendor details
Company policies, strategic plans, and business data
Financial information and operational procedures

Obligations:

Maintain confidentiality of all confidential information.
Use confidential information solely for the performance of duties within the project.
Do not disclose confidential information to any third party without written consent.
Take reasonable precautions to prevent unauthorized access.

non Compete Obligation

During the term of employment and for [e.g., 12 months] after termination, the employee shall not engage in any business, project, or employment that directly competes with the chatbot application or its derivatives.
The employee shall not solicit clients, partners, or employees of the company for a competing venture.

conflict Of Interest

Definition:

A conflict occurs when the employee’s personal, financial, or relational interests may influence, or appear to influence, the employee’s judgment or decisions regarding the project.

Examples:

Personal ownership or investment in a competing AI/chatbot project
Accepting gifts, favors, or benefits from vendors or partners
Engaging in side projects that could compromise confidentiality or business interests

reporting And Compliance

Reporting:

Employees must disclose any potential or actual conflicts to the company promptly.
Management will review disclosures and determine necessary actions, which may include recusal, reassignment, or termination of conflicting activities.

Enforcement:

Violations may result in disciplinary action, including termination, legal action, or financial liability.
Annual conflict-of-interest declarations may be required.