Senior Machine Learning Engineer

About Nexttech

Founded in 2015, Nexttech has built a solid foundation in delivering comprehensive IT solutions tailored to meet diverse client needs. With expertise spanning five key industry sectors—Banking, Energy, Telecom, Automotive and E-commerce & Logistics—we provide nearshore and onshore services designed to drive efficiency and support strategic growth.

Our team supports every phase of the Software Development Life Cycle (SDLC), from developing detailed roadmaps and resolving complex software challenges to ensuring quick time-to-market and optimized ROI.

About the Role:

Behavioral Intelligence Platform – Multi-Model ML System

We are building a machine learning platform where ML is not a feature — it is the product.

Our system consists of seven interdependent ML models, multiple real-time data streams (market data, behavioral events, NLP thesis content, and outcome feedback loops), a structured feature store, and continuous retraining pipelines. The architecture is already defined. Now we need someone who can make it exceptional.

We are looking for a Senior Machine Learning Engineer who can operationalize, refine, and scale a production ML ecosystem that includes:

• Gradient-boosted models (Signal Scorer)
• Variational Autoencoders + GMM clustering (Behavioral Profiler)
• Isolation Forest + CUSUM anomaly detection (Change Detector)
• Fine-tuned FinBERT NLP models (Thesis Quality Analyzer)
• Contextual bandits / reinforcement learning (Strategy Engine)
• DCC-GARCH correlation modeling (Correlation Mapper)
• Genetic algorithm–based ensemble optimization (Composite Strategy Mixer)

This is not a research sandbox. This is a live system with real users, real feedback loops, and real economic impact.

Key Responsibilities:

• Own Model Performance Across the Stack

Improve accuracy, calibration, stability, and robustness of all seven production models.

• Signal Scorer Optimization (Gradient Boosting)

Refine LightGBM models, improve feature engineering, calibration (Platt scaling), and coldstart strategies. Ensure regulatory-grade interpretability.

• Behavioral Modeling (Unsupervised Learning)

Enhance VAE latent space stability, optimize clustering quality (GMM), and prevent profile drift without meaningful signal.

• Anomaly Detection & Behavioral Drift

Reduce false positives in Isolation Forest + CUSUM system while preserving early-warning sensitivity.

•  NLP Model Improvement (FinBERT Fine-Tuning)

Improve thesis quality classification accuracy. Optimize fine-tuning strategy, embedding stability, and label quality separation (TQS vs OQS).

• Contextual Bandit / Reinforcement Learning Optimization

Improve exploration/exploitation trade-offs. Optimize reward shaping using engagement + P&L dual signals. Strengthen counterfactual evaluation.

• Time-Series & Volatility Modeling

Refine correlation modeling (rolling Pearson + DCC-GARCH). Improve regime detection and distribution shift monitoring.

• Genetic Algorithm & Backtesting Framework

Enhance composite strategy optimization robustness. Improve walk-forward validation and survivorship bias correction.

• Data Flywheel Optimization

Strengthen feedback loops: thesis → outcome → retrain. Improve training cadence, drift detection, and model versioning.

• Feature Store & Pipeline Integrity

Ensure training-serving consistency using Feast. Prevent training-serving skew. Optimize large-scale feature computation (Spark).

• Model Governance & Deployment

Maintain MLflow model registry, A/B testing framework, automatic rollback safety, and production monitoring (Evidently + Datadog).

Must-have Skills and Experience:

•  Strong Applied ML Background (5+ years)

Experience deploying and maintaining multiple ML models in production.

• Gradient Boosting & Tabular Modeling

Deep expertise with LightGBM / XGBoost including calibration, feature importance, and imbalanced datasets.

• Unsupervised Learning

Experience with VAEs, clustering (GMM, k-means), latent space modeling, and behavioral segmentation.

• Anomaly Detection

Experience with Isolation Forest, drift detection, statistical process control (CUSUM or similar).

• NLP & Transformer Fine-Tuning

Hands-on experience fine-tuning BERT-family models (FinBERT or similar). Understanding of embedding pipelines.

• Reinforcement Learning / Contextual Bandits

Experience implementing recommendation or policy-learning systems with real feedback signals.

• Time-Series & Financial Modeling

Strong understanding of correlation modeling, volatility clustering, and regime detection.

• Back testing & Validation Rigor

Experience with walk-forward validation, avoiding lookahead bias, survivorship bias correction, and transaction cost modeling.

• Production ML Infrastructure

Experience with:

• Feature stores (Feast or similar)
• MLflow model registry
• Airflow orchestration
• Spark / distributed batch processing
• FastAPI model serving
• AWS (GPU instances, EMR)

• Data Drift & Monitoring

Experience with model monitoring frameworks (Evidently or similar) and automated rollback strategies.

• Strong Python Skills

Production-level coding standards, performance optimization, testing discipline.

Nice to Have:

• Experience in fintech, trading systems, or behavioral finance

• Experience working with large-scale behavioral event datasets

• Knowledge of calibration theory and probabilistic modeling

• Experience optimizing dual-objective reward systems

• Experience with GARCH-family models

• Familiarity with regulatory transparency requirements in financial ML systems

What We Offer

• You are not building a single model — you are evolving a multi-model intelligence ecosystem

• Direct impact on product, retention, and revenue

• ML is the core value driver — not a side feature