Lokesh G

AI/ML Engineer and Data Scientist with 7+ years of experience building machine learning, deep learning, NLP, and production grade data solutions across fraud detection, healthcare, workforce analytics, and financial services. Strong background in developing scalable ML pipelines, deploying models into production, improving model performance through feature engineering and tuning, and partnering with product, engineering, and business teams to deliver measurable results. Experienced in Python, SQL, TensorFlow, PyTorch, Scikit learn, Spark, cloud platforms, and MLOps practices, with a strong focus on solving real business problems using AI at scale.

AI/ML Engineer - Doppel

(2023-08)

• Designed and deployed end-to-end machine learning solutions for high impact business use cases involving fraud detection, entity risk scoring, and behavioral pattern analysis. Built production ready pipelines that handled data ingestion, preprocessing, feature generation, model training, and inference at scale across large and fast changing datasets.
• Developed supervised and unsupervised learning models using Python, Scikit learn, and PyTorch to identify suspicious digital behavior and improve threat detection accuracy. Improved model precision by 18% by redesigning feature engineering workflows and incorporating temporal and behavioral signals into the training data.
• Built NLP-based pipelines to analyze unstructured text, profile metadata, and content level signals for automated classification and risk assessment. Reduced manual review effort by 35% by enabling automated prioritization of high risk cases for internal security and trust teams.
• Created real time and batch inference workflows using FastAPI, Docker, and cloud based services to support model serving in production environments. Reduced average prediction latency by 28% by optimizing model serialization, caching strategies, and endpoint response handling.
• Partnered closely with product managers, data engineers, and security stakeholders to translate ambiguous business problems into measurable AI use cases. Helped move multiple initiatives from proof of concept to production by defining success metrics, validation rules, and rollout strategies early in the project lifecycle.
• Led model monitoring and retraining strategies by tracking concept drift, false positives, feature quality, and inference performance across live systems. Improved post-deployment stability and reduced model degradation issues by building automated alerting and retraining triggers.
• Conducted extensive feature engineering on graph, behavioral, and event-stream datasets to improve classification performance and model robustness. Increased recall on critical fraud scenarios by 21% through richer session level and user feature extraction.
• Built internal experimentation frameworks to compare models, thresholds, and feature sets using offline validation and online testing. Helped the team shorten model experimentation cycles by 30%, allowing faster iteration and more reliable decision making.
• Collaborated with data infrastructure teams to optimize ETL pipelines and training dataset generation for large scale model development. Reduced data preparation time by 40% by automating repeatable transformation steps and improving partition level processing.
• Presented findings, model results, and implementation recommendations to technical and non technical stakeholders through dashboards, reports, and detailed technical walkthroughs. This improved alignment across engineering and business teams and helped secure approval for wider production adoption.

Data Scientist - Traba

(2022-08 - 2023-08)

• Built predictive models to support workforce planning, demand forecasting, and operational decision making using historical labor, engagement, and scheduling data. Helped business teams improve forecast quality and planning efficiency by translating raw data into reliable predictive insights.
• Developed machine learning models for worker matching, shift fill optimization, and performance prediction using structured and behavioral data. Improved matching effectiveness by 17% by refining model features around worker reliability, location patterns, and historical fulfillment behavior.
• Performed exploratory data analysis, feature selection, and statistical testing on high volume operational datasets to identify key performance drivers. Uncovered several high impact trends that influenced staffing strategies and improved resource allocation across business units.
• Designed dashboards and analytical reports in Tableau and Python to communicate trends in worker activity, demand spikes, conversion behavior, and fulfillment outcomes. Reduced reporting turnaround time by 25% by replacing manual reporting processes with automated analytics views.
• Worked closely with product and operations teams to define business KPIs, validate hypotheses, and measure the impact of new marketplace features. Supported more data-driven product decisions by turning broad business questions into testable analytical frameworks.
• Built experimentation and A/B testing frameworks to evaluate product changes, recommendation strategies, and operational process adjustments. Helped the organization improve decision confidence by introducing clearer evaluation metrics and statistically sound measurement methods.
• Automated recurring data science workflows for data extraction, cleaning, model scoring, and reporting using Python and SQL. Reduced manual analytical effort by 30% and improved consistency across recurring decision-support processes.
• Partnered with engineering teams to productionize machine learning and analytics solutions for internal and customer facing systems. Ensured smoother deployment by documenting model assumptions, validation logic, and expected data behavior during implementation.
• Created segmentation and scoring models to better understand worker behavior, retention patterns, and supply quality across regions. Enabled operations teams to focus more effectively on high priority segments and improve engagement strategies.
• Improved model accuracy and generalization by tuning hyperparameters, handling class imbalance, and refining feature pipelines through multiple iteration cycles. Increased prediction performance by 15% on key operational models, resulting in stronger trust from business stakeholders.

ML Engineer - Northwell Health

(2019-08 - 2021-08)

• Developed machine learning solutions for healthcare analytics, clinical risk modeling, and operational process improvement using patient, provider, and administrative datasets. Built scalable workflows that transformed large healthcare datasets into model-ready inputs for predictive analysis and decision support.
• Built classification and risk prediction models to support early identification of patient outcomes, resource planning, and operational bottlenecks. Improved model sensitivity by 16% by redesigning preprocessing steps and introducing clinically relevant derived features into the training pipeline.
• Worked with structured and semi structured healthcare data including EMR related extracts, operational records, and historical patient trends. Standardized preprocessing and validation logic across multiple data sources, which improved data quality and reduced downstream model issues.
• Designed ETL and feature engineering pipelines in Python and Spark to process high volume healthcare data efficiently for training and inference use cases. Reduced end-to-end data preparation time by 32% by automating transformations and optimizing large scale batch processing.
• Collaborated with clinicians, analysts, and cross functional stakeholders to understand use cases and convert healthcare problems into practical ML solutions. Helped bridge technical and domain perspectives by ensuring that models aligned with operational realities and measurable healthcare goals.
• Built model evaluation frameworks using classification metrics, validation checks, and error analysis to improve confidence in production readiness. Reduced false positive rates on selected prediction workflows by 14% through threshold optimization and deeper model diagnostics.
• Supported model deployment and integration efforts by packaging machine learning logic into reusable services and deployment workflows. Improved production stability by documenting dependencies, fallback rules, and monitoring expectations for engineering teams.
• Created visual and statistical analyses to explain model outputs, trends, and risk signals to business users and healthcare stakeholders. Improved adoption by making model results easier to interpret and connect to real operational decisions.
• Contributed to compliance aware data practices by handling sensitive data carefully and working within secure enterprise environments. Helped maintain trust in data science workflows by following approved data usage, validation, and access control procedures.
• Improved team efficiency by documenting reusable code patterns, feature engineering methods, and model development standards for shared use. This supported faster onboarding for new team members and better consistency across model development initiatives.

Python Developer - HDFC Bank

(2017-04 - 2019-08)

• Developed Python based applications and backend utilities to support financial data processing, reporting, and automation needs across internal business functions. Built maintainable and scalable code that improved processing efficiency and reduced manual operational overhead.
• Designed and enhanced data processing scripts for large volume financial datasets using Python, SQL, and shell based workflows. Reduced manual reporting effort by 27% by automating recurring extraction, transformation, and reconciliation tasks.
• Worked on ETL workflows that collected data from multiple upstream systems, transformed it into standardized formats, and loaded it into downstream reporting environments. Improved data consistency and reduced reconciliation issues by enforcing better validation logic during ingestion.
• Built reusable Python modules for reporting, file handling, exception management, and business rule processing across finance related applications. Increased development speed across the team by creating shared components that simplified repeated coding tasks.
• Wrote complex SQL queries, joins, stored procedures, and validation scripts to support business reporting and application logic. Improved query performance in selected workflows by 20% through query tuning, indexing awareness, and better transformation design.
• Collaborated with analysts, testers, and business users to gather requirements, troubleshoot production issues, and deliver enhancements to existing applications. Improved delivery quality by making sure technical changes reflected actual user needs and operational expectations.
• Supported application maintenance and production issue resolution by debugging Python jobs, validating outputs, and fixing logic related defects. Reduced recurring production incidents by improving error handling, logging, and edge case validation across key workflows.
• Automated file based and database driven processes that previously required significant manual intervention from operations teams. Helped cut turnaround times for selected reporting cycles and improved consistency in time sensitive financial processes.
• Participated in code reviews, testing, release support, and documentation efforts to maintain quality across ongoing development work. Helped strengthen code reliability and knowledge sharing by contributing to better development standards and clearer technical documentation.
• Built a strong foundation in software engineering, data handling, and enterprise application support in a highly regulated financial environment. This experience helped shape later work in machine learning and data science by combining technical depth with business discipline.

Master's - Data Science - Rowan University

Bachelor's - Mechanical Engineering - KL University