Machine Learning Engineer - Python/PyTorch

MACHINE-LEARNING ENGINEER.

About Us

Datacultr is a global Digital Operating System for Risk Management and Debt Recovery, we drive Collection Efficiencies, Reduce Delinquencies and Non-Performing Loans (NPL's).

Datacultr is a Digital-Only provider of Consumer Engagement, Recovery and Collection Solutions, helping Consumer Lending, Retail, Telecom and Fintech Organizations to expand and grow their business in the under-penetrated New to Credit and Thin File Segments.

We are helping millions of new to credit consumers, across emerging markets, access formal credit and begin theirjourney towards financialhealth.

We have clients acrossIndia, South Asia, South East Asia, Africa and LATAM.

Datacultr is headquartered in Dubai, with offices in Abu Dhabi, Singapore, Ho Chi Minh City, Nairobi, and Mexico City; and our Development Center is located out of Gurugram, India.

Organization's Growth Plan

Datacultr's vision is to enable convenient financing opportunities for consumers, entrepreneurs and small merchants, helping them combat the Socio-economic problems this segment faces due to restricted access to financing.

We are on a mission to enable 35 million unbanked& under-served people,access financial services by the end of 2026.

Position Overview

We're looking for an experienced Machine Learning Engineer to design, deploy, and scale production-grade ML systems.

You'll work on high-impact projects involving deep learning, NLP, and real-time data processing-owning everything from model development to deployment and monitoring while collaborating with cross-functional teams to deliver impactful, production-ready solutions.

Core Responsibilities

Representation & Embedding Layer :

• Evaluate, fine-tune, and deploy multilingual embedding models (e.g, OpenAI text-embedding-3, Sentence-T5, Cohere, or in-house MiniLM) on AWS GPU or serverless endpoints.
• Implement device-level aggregation to produce stable vectors for downstream clustering.
  
  

Cohort Discovery Services

• Build scalable clustering workflows in Spark/Flink or Python on Airflow.
• Serve cluster IDs & metadata via feature store / real-time API for consumption.
  
  

MLOps & Observability

• Own CI/CD for model training & deployment.
• Instrument latency, drift, bias, and cost dashboards; automate rollback policies.
  
  

Experimentation & Optimisation

• Run A/B and multivariate tests comparing embedding cohorts against legacy segmentation; analyse lift in repayment or engagement.
• Iterate on quantisation, distillation, and batching to hit strict cost-latency SLAs.
  
  

Collaboration & Knowledge-sharing

• Work hand-in-hand with Product & Data Strategy to translate cohort insights into actionable product features.
  
  

Key Requirements

• 5-8 years of hands-on ML engineering / NLP experience; at least 2 years deploying transformer-based models in production.
• Demonstrated ownership of pipelines processing - 100 million events per month.
• Deep proficiency in Python, PyTorch/TensorFlow, Hugging Face ecosystem, and SQL on cloud warehouses.
• Familiar with vector databases and RAG architectures.
• Working knowledge of credit-risk or high-volume messaging platforms is a plus.
• Degree in CS, EE, Statistics, or related;.
  
  

Tech Stack You'll Drive

• Model & Serving - PyTorch, Hugging Face, Triton, BentoML.
• Data & Orchestration - Airflow, Spark/Flink, Kafka.
• Vector & Storage - Qdrant/Weaviate, S3/GCS, Parquet/Iceberg.
• Cloud & Infra - AWS (EKS, SageMaker).
• Monitoring - Prometheus, Loki, Grafana.
  
  

What We Offer

• Opportunity to shape the future of unsecured lending in emerging markets.
• Competitive compensation package.
• Professional development and growth opportunities.
• Collaborative, innovation-focused work environment.
• Comprehensive health and wellness benefits.
  
  

Location & Work Model :

• Immediate joining possible.
• Work From Office only.
• Based in Gurugram, Sector 65.
  
  

(ref:hirist.tech)