AI Engineer Lead

Allstate

20h

About The Position

At Allstate Investments, we are seeking an AI Developer who is a fast learner and rapid problem solver, capable of quickly understanding new domains, technologies, and business problems, and translating them into high impact proof-of-concept (PoC) solutions. This role requires the ability to move quickly from idea to working prototypes, validate value early, and then guide successful PoCs into scalable, production-ready AI solutions. As a member of the Investments Technology group, your work directly supports an $80bn portfolio and an organization of 300+ experts committed to excellence, innovation, and the highest ethical standards. You will be a hands-on developer – pairing daily with other engineers and guiding adoption of LLMs, RAG, vector stores, and agentic frameworks. You’ll partner closely with product and platform leaders to build AI systems.

Requirements

Minimum of 5 years of python development
Minimum of 3 years of experience working on AI projects, with a focus on productizing AI/ML solutions.
Deep understanding of machine leaning fundamentals and ability to apply them in developing and implementing AI/ML solutions.
Experience with GenAI technologies and techniques (e.g., RAG, fine-tuning, vector stores, Agentic frameworks).
Data-driven decision making, combining robust analytical thinking with sound business judgment.
Excellent communication with both technical and non-technical team members.
Comfort with hands-on involvement in AI/ML experimentation and model development.
Solid grasp of software development principles, including design patterns, testing, and version control such as Git.
Programming & Engineering Excellence: Python (advanced), APIs, pipelines, and AI orchestration
LLM & GenAI Engineering: Hands-on experience with LLMs (preferably GPT models)
Prompting patterns and structured outputs
Model evaluation including accuracy/faithfulness checks, hallucination mitigation, regression testing
Safety/guardrails patterns appropriate for enterprise use
Demonstrated ability to leverage GitHub Copilot for code acceleration: Rapid scaffolding of code, services, APIs, and development
Data Skills — SQL + SQL Server, Microsoft Fabric(optional)
SQL skills: Query authoring, joins, indexing awareness
Ability to design data access patterns for AI
LLM & GenAI Engineering: Hands-on experience with LLMs (commercial)
Prompting patterns and structured outputs
Strong experience implementing RAG architecture with ingestion pipelines, chunking strategies
Vector database / vector search experience with Indexing, similarity search, metadata filtering
Experience with MCP patterns, Standardizing tool/context access for models and agent runtimes
Designing MCP servers/tool endpoints and integrating them into LLM apps
Experience implementing agentic workflows, A2A patterns such as multi-agent collaboration (planner/executor/reviewer/retriever roles)
Cloud: MS Azure Fabric preferred
Observability & Production Support: Experience operating and supporting AI-enabled services in production: Logging, metrics, tracing (APM tools such as Datadog)
Model/prompt monitoring, drift signals, quality regression detection

Nice To Haves

Java/ReactJS (optional)
Inquisitive nature, with a deep understanding of both technical and business aspects.

Responsibilities

Collaborate with a highly talented and diverse team of professionals
Continuously expand your skill set by learning new technologies, tools, and methodologies.
Proactively identify opportunities for innovation, with the autonomy to rapidly explore, prototype, and validate new ideas.
Play a key role in the design, development, and deployment of AI/ML solutions.
Lead the transition of AI/ML initiatives from proof of concept to scalable, production-ready solutions.
Establish and socialize strong AI engineering fundamentals, design patterns, code quality standards.
Participate in projects, demonstrating technical expertise and a hands-on approach in experimentation and productization.
Partner with product stakeholders to iterate quickly on PoCs, incorporating feedback and refining solutions in short cycles.