AI Software Engineer

Job Duties:

Research, Design and implement an interactive AI/ML-driven system for generating nondeterministic automated processes from natural language input, the system is to be accessible to end users through natural language-based Agents. Establish the system atop a reconstructive memory which is fluid, shaped by semantic meaning, enabling adaptive interpretation of user needs and intelligent process optimization (continuous process improvement).

‍

Design and implementation of a system which identifies complex issues and generates precise diagnoses and automates problem resolution through sophisticated text and image pattern recognition and contextual understanding based on a heterogeneous knowledge base augmented with LLMs. The system must possess robust troubleshooting capabilities, offering insightful guidance and leading users through systematic problem-solving pathways. Design and build a system that self-corrects and validates the accuracy and viability of its outputs from a Large Language Model (LLM). Using multi-metric scoring to assess the accuracy, contextual relevance, and factual consistency of outputs across generation, classification, and retrieval tasks. Implement scoring pipelines that combine Answer Relevance, Contextual Relevance, Groundedness, and BERTScore to generate rationale-aligned quality metrics. Use these evaluations to train reward models via reinforcement learning and integrate them into production workflows for fine-tuning, adaptive calibration, and continuous model optimization. 20% domestic travel required. Hybrid work policy w/in commuting distance

Job Requirements:

Master’s degree or foreign equivalent in Computer Science or a related field, and 5 years of experience as a software engineer for machine learning and large-scale natural language processing systems.

Requires 3 years of experience in the following:

• Building multi-model enterprise grade agentic systems; Designing reconstructive memory systems characterized by dynamic, context-sensitive, and self-forming associative relationships.
• Dynamical (non-deterministic) non-Markovian code generation; Creating patentable or potentially patentable IP.
• Designing runtime code generation mechanisms; designing & operationalizing evaluation frameworks for large language models (LLMs) across generation, classification, and retrieval tasks.
• Designing runtime execution engines which generate or modify its code based on non-Markovian long-term dependencies and conditions encountered during execution, probabilistically choosing and generating the best course of action.
• Building multi-metric LLM scoring pipelines including Answer Relevance, Contextual Precision, Groundedness, and BERTScore to generate reward signals for reinforcement learning and continuous optimization.
• Python and ML libraries including scikit-learn, TensorFlow, and related frameworks for model development and data processing.
• Building scalable retrieval systems using vector databases, semantic embeddings, metadata-based filtering, and contextual reranking techniques.
• Containerized deployment and CI/CD pipelines using Docker, GitHub Actions, and Azure DevOps for ML system integration and delivery.

Requires 2 years of experience in the following:

• Hands-on experience with LLM platforms and prompt engineering.
• Designing, implementing, and managing agent-based systems and workflow automation using orchestration frameworks and modular control platforms.
• Building scalable retrieval systems using vector databases, semantic embeddings, metadata-based filtering, and contextual reranking techniques.
• Evaluation & RLHF frameworks and equivalent tools for assessing model performance, reward modeling, and fine-tuning.
• Model Context Protocol (MCP) to manage contextual memory, input routing, and task coordination across multi-agent workflows, enabling modular, explainable, and stateful LLM-driven systems.