Quantitative Developer

HarbourVest Partners•Boston, MA

1d•$140,000 - $155,000•Onsite

About The Position

Seated within our Quantitative Investment Science group, this position turns machine learning, applied AI, and agentic workflow capabilities into reliable investment workflow software. This is a software engineering role first: you will write production Python, work deeply with data, build model pipelines and evaluation frameworks, and integrate AI-driven capabilities into the tools investment teams use every day. The role is ideal for a practical machine learning engineer who wants to build trusted, auditable systems for high-value quantitative and private markets workflows. The ideal candidate is someone who has: Strong software engineering fundamentals and a production-oriented machine learning mindset A practical interest in using ML and agentic AI to improve investment research, data quality, decision support, and workflow scale Healthy skepticism about model outputs, with strong instincts for evaluation, backtesting, monitoring, and human review Comfort turning ambiguous analytical workflows into measurable, maintainable production systems Strong collaboration skills across quant developers, data engineering, product, and investment stakeholders Curiosity about finance, private markets, and the data problems behind investment decision-making

Requirements

Strong proficiency in Python and modern software engineering practices
Experience with applied machine learning, including feature engineering, model training, evaluation, inference, and monitoring
Ability to learn and apply the right ML, statistical, and data engineering tools for the problem, with sound judgment around model choice, data representation, reproducibility, and production constraints
Strong SQL skills and comfort designing data models for analytical or product-facing systems
Experience building production services, APIs, batch jobs, queues, or scheduled pipelines around data-intensive workflows
Practical experience with embeddings, semantic search, ranking, recommendation systems, information extraction, agentic AI systems, or LLM-enabled workflows
Familiarity with agent patterns such as tool use, retrieval-augmented generation, planning, memory, workflow orchestration, and structured human review
Strong testing habits and ability to debug model behavior using real data, logs, metrics, and user feedback
Ability to explain model behavior, data limitations, quality tradeoffs, and operational risk to technical and non-technical partners
Familiarity with cloud platforms, containerized development, CI/CD, observability, and secure production deployment patterns

Nice To Haves

Experience with financial data, time series data, private markets workflows, vector databases, agent frameworks, unstructured data processing, feature stores, model registries, or multi-tenant enterprise systems

Responsibilities

Build and productionize ML models, feature pipelines, and inference workflows for QIS applications
Develop semantic matching, ranking, recommendation, and peer-selection systems for funds, managers, deals, companies, and comparable opportunities
Build unstructured data intelligence, classification, enrichment, and AI-assisted review workflows for complex internal materials and operational datasets
Design agentic AI workflows that can plan multi-step analyses, call internal tools, retrieve relevant context, and produce traceable recommendations for human review
Create evaluation frameworks for AI agents, including task success metrics, regression suites, prompt/version tracking, guardrail tests, and failure-mode analysis
Create model evaluation harnesses, benchmark datasets, backtests, monitoring, drift detection, and quality gates so ML outputs can be measured and trusted
Integrate embeddings, retrieval, model-serving APIs, agent orchestration, batch jobs, and human-in-the-loop review controls into existing QIS tools
Partner with data and platform engineers to make ML workflows repeatable, observable, secure, and easy to operate
Establish practical MLOps patterns for experiment tracking, model versioning, deployment, rollback, audit trails, and production support
Translate investment workflow needs into pragmatic ML solutions while being clear about limitations, confidence, and operational risk