Machine Learning Engineer

Spotlight Partners; Michigan; Hybrid; USD 120,000 - 170,000; Full time

Benefits offered by this job

Hybrid flexibility (Troy/Birmingham,MI

Real ownership of the retrieval/model

Job summary

Spotlight Partners is a privately held SAP consulting firm at the intersection of SAP Cloud ERP and AI transformation, helping clients accelerate toward streamlined operations and untapped potential. We specialize in building investment-grade business cases and delivering Beacon, our AI platform, to turn engineering proofs into saleable software.

We pride ourselves on a balance of business operations and technology, with ownership and quality as core values, and we seek someone to translate

Qualifications

• Strong Python skills with experience in PyTorch, Hugging Face Transformers, or scikit-learn.
• Experience with LangChain, LlamaIndex, or similar orchestration frameworks.
• Hands-on experience with vector stores (pgvector, Pinecone, Weaviate, or similar).
• Solid grounding in transformer architecture, embeddings, and RAG design, including hybrid search and reranking.
• Judgment for when to fine-tune, when to retrieve, and when to simply prompt — plus familiarity with quantization, distillation, and drift detection.
• Comfort with pandas, NumPy, SQL, and experiment tracking (MLflow, Weights & Biases, or similar).
• Nice to have: Professional ML Engineer, or Databricks ML Professional.

Responsibilities

• Design and tune retrieval, ranking, and reranking
• Own our embedding and chunking strategy
• Fine-tune or distill models, but only where it measurably beats prompting
• Benchmark model options for cost and latency
• Build feature and training pipelines for predictive components
• Partner with our evaluation engineer on measurement design

Skills

Python

PyTorch

Hugging Face

scikit-learn

LangChain

LlamaIndex

Vector stores

Transformer basics

Embeddings

RAG design

Hybrid search

Fine-tuning judgment

Pandas

NumPy

SQL

MLflow

Weights & Biases

Tools

pgvector

Pinecone

Weaviate

Hugging Face Transformers

LangChain

LlamaIndex

MLflow

Weights & Biases

Job description

Spotlight Partners is a privately held SAP consulting firm that lives at the intersection of SAP Cloud ERP and AI transformation — helping clients accelerate toward streamlined operations and untapped potential.

We specialize in two things: creating investment-grade business cases that articulate exactly the expected return on an SAP engagement, and taking a proprietary approach to developing supplemental AI within and outside the SAP ecosystem. That approach is Beacon, our AI platform — and we need someone to turn what our engineering team proves out into software we can actually sell and support.

Spotlight Partners is a uniquely experienced and curated group of talent built to deliver balance across

business operations and technology, without the red tape and bureaucracy of other consulting firms.

None of this works without the right people, and we've been deliberate about who's in the room. Spotlight Partners asks a lot — sound judgment, real ownership, and a genuine commitment to clients over convenience. What you get in return is a team that makes you sharper: people who know the work, say what they think, and hold the line on quality even when it's inconvenient. It's demanding. It's also where people do the best work of their careers.

The Role

You’ll make the models and retrieval behind Beacon accurate, fast, and cheap. That means rigorous

benchmarking before optimization, and being honest about when fine-tuning is worth it versus when a better prompt or retrieval fix would do the job just as well.

What You’ll Do

• Design and tune retrieval, ranking, and reranking
• Own our embedding and chunking strategy
• Fine-tune or distill models, but only where it measurably beats prompting
• Benchmark model options for cost and latency
• Build feature and training pipelines for predictive components
• Partner with our evaluation engineer on measurement design

What You’ll Bring

• Strong Python skills, with experience in PyTorch, Hugging Face Transformers, or scikit-learn
• Experience with LangChain, LlamaIndex, or similar orchestration frameworks
• Hands-on experience with vector stores (pgvector, Pinecone, Weaviate, or similar)
• Solid grounding in transformer architecture, embeddings, and RAG design, including hybrid search and reranking
• Judgment for when to fine-tune, when to retrieve, and when to simply prompt — plus familiarity with
• quantization, distillation, and drift detection
• Comfort with pandas, NumPy, SQL, and experiment tracking (MLflow, Weights & Biases, or similar)
• A measure-before-you-optimize mindset, and the ability to explain a tradeoff in terms a client would
• recognize

Nice to Have

• Professional ML Engineer, or Databricks ML Professional

What We Offer

• Real ownership of the retrieval and model layer behind a revenue-generating AI product
• A team that measures results rather than chasing demo-only wins
• Hybrid flexibility (Troy/Birmingham, MI area)
• Competitive compensation, commensurate with experience