Machine Learning Engineer Intern

vaisalaoyj

2d•Hybrid

About The Position

Vaisala is seeking a Machine Learning Engineer Intern to take existing RWIS predictive station recovery work from prototype into a robust, operational system. Over an initial 3 to 6 month period, the intern will harden and extend existing models, integrate them with RWS200 production data, and deliver actionable alerts that reduce unnecessary truck rolls and improve station uptime across 84 PennDOT sites. This will be a full-time role based in their Louisville, Colorado office on a hybrid work schedule.

Requirements

Master's in Computer Science, Data Science or a related quantitative field.
Strong background in machine learning, with experience in classification and/or time‑series/event prediction.
Hands‑on experience with sensor or telemetry data, including sliding‑window feature engineering and noisy data.
Proficiency in Python and common ML/data tools (e.g., pandas, scikit‑learn; deep learning frameworks a plus).
Demonstrated experience shipping ML models into production (batch or real‑time), including monitoring and iteration.
Familiarity with NLP for short text classification (e.g., ticket descriptions, subject lines) or willingness to quickly ramp up.
Ability to work cross‑functionally with operations, product, and engineering stakeholders.
Bachelor’s or Master’s in Computer Science, Data Science, or a related quantitative field, or equivalent practical experience.
Currently eligible to work in the U.S.

Nice To Haves

Experience with industrial IoT, environmental sensing, or monitoring systems (e.g., weather, energy, industrial equipment).
Experience with cost‑sensitive learning, imbalanced datasets, and threshold tuning driven by real‑world operational costs.
Familiarity with MLOps practices and tooling (experiment tracking, CI/CD for ML, model registries).
Experience with the cloud and data stack similar to Vaisala’s (you can specify AWS/Azure/GCP and any internal technologies here).

Responsibilities

Own the end‑to‑end lifecycle of ML models for RWIS predictive station recovery, from exploration and feature engineering through deployment and monitoring.
Engineer robust data pipelines over multi‑year sensor telemetry and CRM case data, with an emphasis on time‑window features and noisy/partial data handling.
Build, train, and iterate on models for: urgency classification, self‑recovery prediction, failure‑type classification from case text, and sensor‑level degradation/gap detection.
Translate operational and business costs (truck rolls, downtime, false alarms) into cost‑aware evaluation metrics and decision thresholds.
Collaborate closely with operations/support teams to align alerts and routing with existing workflows and constraints.
Implement production integrations, logging, observability, and model performance monitoring.
Document architecture, assumptions, and operational procedures to support future scaling and handoff.