Senior AI/ML Engineer, Gesture

• Do not apply unless you can honestly say yes to all of the following:
• You have 5+ years of hands-on ML engineering or applied AI experience with models running in production, not just in research or evaluation
• You have personally owned the full ML lifecycle: data, features, training, deployment, monitoring, and retraining
• You are proficient in Python,SQL and have production experience with ML frameworks (PyTorch, TensorFlow, scikit-learn, XGBoost, or equivalent)
• You have designed and deployed LLM-powered features: RAG pipelines, fine-tuning workflows, vector embeddings, or semantic search in production
• You understand data modeling, feature engineering, and pipeline design for both real-time and batch inference at scale
• You have experience with ML orchestration tools (Airflow, Prefect, Kubeflow, or similar)
• You have built and maintained production data pipelines from external sources - not just consumed them; you've handled API integration failures, schema drift, and data quality issues upstream of any model
• You have experience with event-driven pipeline architecture: Pub/Sub, Dataflow, or equivalent
• You write clean, production-grade code -- not prototype code that needs to be rewritten before it ships
• You are comfortable with cloud-native ML infrastructure (GCP preferred: Vertex AI, BigQuery, Cloud Run)
• Designing and building scalable systems for model development, training, and deployment, managing large-scale data pipelines and distributed compute environments
• You can make model architecture decisions under ambiguity and course-correct fast when the data tells you to
• You communicate technical concepts to non-technical stakeholders without losing accuracy
• HOW YOU OPERATE:
• Systems Thinking: You see the full stack, from raw signal to user-facing outcome, and you build for that entire path -- not just your slice of it
• Production Orientation: A model that doesn't work reliably in production is a model that doesn't work. You never lose sight of that
• Bias for Action: You build, test, and ship with incomplete information and improve from real-world feedback
• Accountability: You hold yourself to the performance of your models in production, not just at evaluation time
• Ego-Free: Your success is measured by what the Intelligence Engine does for the product and the business, not your own visibility within the team
• Range: Scoring architecture in the morning, LLM evaluation at noon, pipeline debugging by 3pm. You navigate all of it
• Rigor: You care about statistical validity, model fairness, and production reliability not because someone is checking, but because the alternative is unacceptable

Desirable

• Department: Engineering & Technology
• Function: Artificial Intelligence, Machine Learning & Data Science
• Reports To: VP of Engineering
• Direct Reports: Scope scales with organizational growth
• Travel: As required
• This is a front-row seat to the future of connection. If you want to help build something that's changing how the world interacts, welcome to Gesture
• Gesture is a late-stage, venture-backed technology company in full hyper-scale mode
• We are past the experimental stage
• The product works, the market is real, and the Intelligence Engine is what powers our differentiation
• What we need now is an engineer who can build, ship, and own AI and ML systems that perform in production -- not in a notebook, not in a pilot, and not behind a research wall
• Someone who treats model quality the same way a revenue operator treats a missed number: personally
• The Senior AI/ML & Data Engineer Senior AI/ML Engineer is the technical backbone of the Intelligence Engine
• You sit at the intersection of data science, machine learning, and software engineering -- designing the systems that score behavior, surface insight, and activate outcomes across every surface of the Gesture platform
• Let's be direct about the environment: this is not a place for researchers who hand off to "implementation teams." It is not a place for engineers who prototype endlessly without shipping, who treat model accuracy as someone else's problem, or who need perfect data to start building
• There are no layers to hide behind, no slow approval chains, and no tolerance for the gap between what models do in testing and what they do in production
• If your instinct when a model underperforms is to document it in a ticket, stop reading here
• If your instinct is to fix it, keep going
• You will personally own:
• The Intelligence Engine -- Scoring & Activation -- The behavioral scoring algorithms that power user-facing activation across B2C and B2B; you own the logic, the lift, and the outcomes
• ML Pipeline Architecture -- The full lifecycle: data ingestion, feature stores, model training, evaluation, deployment, monitoring, and retraining triggers
• LLM Integration & Optimization -- Integration, fine-tuning, and production deployment of large language models for contextual inference, personalization, and behavioral pattern recognition
• Behavioral Signal Processing -- Extraction of meaningful features from structured and unstructured behavioral data; you define what signals matter and why
• Model Evaluation & Governance -- Evaluation frameworks, A/B testing infrastructure, bias detection, and drift monitoring; model integrity in production is non-negotiable
• Data Infrastructure -- Scalable pipeline and schema design that supports real-time and batch inference, built in collaboration with engineering
• Research to Production -- The full path from experimentation to deployed, monitored, and maintained production systems; nothing lives in a notebook forever
• Cross-Functional Technical Partnership -- Direct collaboration with product, engineering, and leadership to translate business outcomes into machine learning problem definitions
• You will own the intelligence layer of the product. You will not "support" it
• Let's be clear:
• This is not a research role with no path to production
• This is not a data analyst position
• This is not a "build models and hand off to engineering" job
• This is not a role for engineers who treat deployment, monitoring, and retraining as someone else's problem
• This is not a role for people who need clean data, perfect infrastructure, or a fully staffed ML team to be effective
• If you have never owned a model in production -- including when it breaks, drifts, or underperforms -- this will be painful
• You own:
• Model performance in production, not just in evaluation; lift, precision, recall, and business impact are your metrics
• End-to-end ML pipeline reliability: data in, scored output out, on time, every time
• Speed from experimentation to production; the faster a good idea gets to users, the better
• Signal quality and feature engineering that makes models better than the data suggests they should be
• Model governance: version control, drift detection, bias auditing, and documentation that holds up to scrutiny
• Infrastructure decisions that scale without requiring a rewrite every six months
• Cross-functional clarity: engineering, product, and leadership should always understand what the models do, why they do it, and how to measure whether it's working
• The technical roadmap for the Intelligence Engine; where we are, where we're going, and what's blocking progress
• When the Intelligence Engine performs and the product is smarter because of your work, you get credit. When model quality degrades and no one caught it, that falls on you
• I. The Intelligence Engine -- Scoring & Behavioral Intelligence:
• Design, build, and continuously improve the behavioral scoring system that powers Gesture's core activation model across B2C and B2B
• Own signal selection, feature engineering, and model architecture decisions for all scoring use cases
• Define and enforce evaluation standards for every model that touches a user-facing surface
• Build the feedback loops that keep models improving as behavior data compounds
• II. ML Pipeline & Infrastructure:
• Own the full ML lifecycle from raw data to deployed inference: ingestion, transformation, training, evaluation, deployment, and monitoring
• This includes owning the raw data layer - you are not handed clean features. You build and maintain ETL/ELT pipelines from third-party data sources (APIs, webhooks, flat files), handle auth, rate limits, and schema drift, and design event-driven ingestion using Pub/Sub and Dataflow for real-time signal processing
• Design pipeline architecture for both real-time and batch inference use cases
• Build and maintain the feature store, experiment tracking, and model registry that make the ML organization productive at scale
• Partner with engineering to ensure ML infrastructure is production-grade, not prototype-grade
• III. LLM Integration & Applied AI:
• Integrate and optimize large language models for contextual inference, personalization, and behavioral pattern detection across Gesture's platform
• Evaluate and implement retrieval-augmented generation, fine-tuning, and embedding workflows based on product requirements
• Make rigorous tradeoff decisions on model selection: cost, latency, accuracy, and reliability in production
• Build and maintain vector search and semantic retrieval capabilities that power personalization at scale
• IV. Data Science & Experimentation:
• Design and execute experiments that produce statistically valid, business-relevant results
• Own A/B testing infrastructure and experiment design standards across the ML team
• Develop causal inference and multi-armed bandit frameworks where appropriate to accelerate learning velocity
• Translate experiment results into model improvements, product decisions, and actionable recommendations for leadership
• V. Model Governance & Quality:
• Build and maintain model monitoring dashboards that surface degradation, drift, and anomalies before they become product issues
• Own bias auditing and fairness evaluation across all scoring and classification systems
• Maintain documentation, versioning, and rollback capability for all production models
• Establish and enforce model quality standards across the team as the ML function scales
• VI. Cross-Functional Technical Leadership:
• Translate business problems into machine learning problem definitions with clear success criteria
• Communicate model behavior, performance, and limitations to non-technical stakeholders without losing precision
• Partner with product and engineering to sequence ML work against business priorities
• Build the technical foundation that allows a growing ML team to operate without creating chaos

• Interested candidates should submit the following:
• A resume outlining relevant experience
• A brief cover letter highlighting your key skills, achievements, and what excites you about this opportunity
• Any project work, portfolio samples, GitHub repositories, or published research that demonstrate your passion and abilities