Note: The job is a remote job and is open to candidates in USA. PAR Technology is hiring an ML Ops Engineer for Bridg, partnering with its data science team to productionize an identity resolution model. The role builds and maintains automated pipelines, tests, CI/CD, and observability to keep match data reliable at scale.
Responsibilities
- Your focus is to help take their work from notebooks to monitored, automated production — building and maintaining the pipelines, tests, and observability that keep our match data reliable at scale
- You will spend most of your week writing code: building pipelines, tuning Spark jobs, wiring up CI/CD, and leaving the system better tested and better documented than you found it
- Daily, fluent use of AI coding assistants (Claude Code, Cursor, Copilot, or equivalent) as a core part of your workflow
- Good judgment about when to delegate to AI vs. write by hand — scaffolding tests, drafting refactors, generating boilerplate and IaC, and explaining unfamiliar code, while writing the tricky parts yourself
- Holding AI-generated code to the same review bar as your own: you read it, test it, and are accountable for it
- Using AI to accelerate non-coding work too — drafting design docs, summarizing logs and runs, exploring datasets, and generating eval cases
Skills
- 2–4 years of experience building production data or ML systems, with a Bachelor's degree in Computer Science, Data Engineering, or a related field (or equivalent practical experience)
- A highly motivated, curious, and capable learner with a strong sense of ownership and accountability
- A strong communicator and collaborator who works well across cross-functional teams
- Comfortable in a fast-moving environment — able to shift priorities, navigate ambiguity, and support a customer-facing system
- Experience building and maintaining automated, tested ML or data pipelines in partnership with other teams (you don't need to be a from-scratch model builder)
- Experience optimizing CI/CD for data/ML: Git-based workflows, automated unit and integration tests (pytest or similar), and branch-based promotion through dev/stage/prod
- Workflow orchestration with Airflow, Step Functions, or Databricks Workflows, including retries, backfills, and basic SLA monitoring and alerting
- Familiarity with experiment tracking and model registries (MLflow or equivalent) for lineage and stage transitions
- Exposure to monitoring and observability — data-quality and schema-drift checks (such as Great Expectations, Deequ, or Datadog) and awareness of input/output distribution drift
- Secrets management, and reproducible Python environments (Docker, Poetry/uv/Conda)
- Production experience on AWS (S3, EMR, Glue, Step Functions, Lambda, IAM, CloudWatch)
- Working knowledge of Spark performance — comfortable diagnosing and addressing common issues like skew, shuffle, and spill, with sensible partitioning and join strategies
- Experience serving processed outputs to downstream systems (e.g., Redshift for BI, activation, or attribution)
- A strong technical partner to data science (you're not the lead data scientist on IDR — that's covered)
- Able to run EDA and data-quality investigations: profiling datasets, surfacing anomalies, and communicating findings clearly
- Experience building and maintaining feature pipelines and keeping training and serving consistent
- A solid applied-stats foundation (probability, hypothesis testing, sampling, evaluation metrics) — enough to support champion/challenger evaluation and flag when something looks off
- Comfortable reading ML code you didn't write and asking good questions; fluent across the Python DS stack (pandas, scikit-learn, PySpark)
- Interest in how LLM-based agents consume structured data — tool definitions, retrieval patterns, and pre-computed signals vs. raw data
- Willingness to build lightweight interfaces that make IDR outputs queryable by agent workflows (MCP servers, REST/GraphQL endpoints, or structured feeds)
- Good judgment about data contracts — what to expose, at what granularity, and how to keep it maintainable
- Daily, fluent use of AI coding assistants (Claude Code, Cursor, Copilot, or equivalent) as a core part of your workflow
- Good judgment about when to delegate to AI vs. write by hand — scaffolding tests, drafting refactors, generating boilerplate and IaC, and explaining unfamiliar code, while writing the tricky parts yourself
- Holding AI-generated code to the same review bar as your own: you read it, test it, and are accountable for it
- Using AI to accelerate non-coding work too — drafting design docs, summarizing logs and runs, exploring datasets, and generating eval cases
- Infrastructure as code (Terraform preferred)
- Experience with agentic frameworks (LangChain, LlamaIndex, or similar) is a plus
- Exposure to PII-based record linkage, probabilistic matching methodology, or graph/clustering-based entity resolution (e.g., connected components)
- Awareness of blocking strategies and the recall/runtime tradeoffs they drive at large record volumes
- Comfort working carefully with PII-bearing datasets (our internal processes cover the privacy and compliance side)
- Cost-optimization wins on EMR, AWS, Databricks, or equivalent Dev environment
Benefits
- Bonus
- Hybrid work arrangement
Company Overview
Company H1B Sponsorship