Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I’m Harika Jeksani, an AI/ML Engineer with around 4 years of experience delivering production-grade ML and Generative AI solutions across finance and retail. I specialize in Python ecosystems (Scikit-Learn, PyTorch, TensorFlow) and have hands-on experience building NLP and LLM-based applications using Hugging Face, LangChain, and RAG pipelines. I enjoy turning complex data into practical, business-focused AI solutions that scale. I’ve led the end-to-end lifecycle from data processing and feature engineering to model deployment and monitoring, with strong ML Ops practices (MLflow, Docker, REST APIs, and AWS). My work consistently improves efficiency and speeds up processing for high-volume systems handling millions of records.…I’m Harika Jeksani, an AI/ML Engineer with around 4 years of experience delivering production-grade ML and Generative AI solutions across finance and retail. I specialize in Python ecosystems (Scikit-Learn, PyTorch, TensorFlow) and have hands-on experience building NLP and LLM-based applications using Hugging Face, LangChain, and RAG pipelines. I enjoy turning complex data into practical, business-focused AI solutions that scale. I’ve led the end-to-end lifecycle from data processing and feature engineering to model deployment and monitoring, with strong ML Ops practices (MLflow, Docker, REST APIs, and AWS). My work consistently improves efficiency and speeds up processing for high-volume systems handling millions of records.

Harika Jeksani

AI Engineer, Data Scientist, Full Stack Developer, +2





I’m Harika Jeksani, an AI/ML Engineer with around 4 years of experience delivering production-grade ML and Generative AI solutions across finance and retail. I specialize in Python ecosystems (Scikit-Learn, PyTorch, TensorFlow) and have hands-on experience building NLP and LLM-based applications using Hugging Face, LangChain, and RAG pipelines. I enjoy turning complex data into practical, business-focused AI solutions that scale. I’ve led the end-to-end lifecycle from data processing and feature engineering to model deployment and monitoring, with strong ML Ops practices (MLflow, Docker, REST APIs, and AWS). My work consistently improves efficiency and speeds up processing for high-volume systems handling millions of records.…I’m Harika Jeksani, an AI/ML Engineer with around 4 years of experience delivering production-grade ML and Generative AI solutions across finance and retail. I specialize in Python ecosystems (Scikit-Learn, PyTorch, TensorFlow) and have hands-on experience building NLP and LLM-based applications using Hugging Face, LangChain, and RAG pipelines. I enjoy turning complex data into practical, business-focused AI solutions that scale. I’ve led the end-to-end lifecycle from data processing and feature engineering to model deployment and monitoring, with strong ML Ops practices (MLflow, Docker, REST APIs, and AWS). My work consistently improves efficiency and speeds up processing for high-volume systems handling millions of records.

Available to hire

I’m Harika Jeksani, an AI/ML Engineer with around 4 years of experience delivering production-grade ML and Generative AI solutions across finance and retail. I specialize in Python ecosystems (Scikit-Learn, PyTorch, TensorFlow) and have hands-on experience building NLP and LLM-based applications using Hugging Face, LangChain, and RAG pipelines. I enjoy turning complex data into practical, business-focused AI solutions that scale.

I’ve led the end-to-end lifecycle from data processing and feature engineering to model deployment and monitoring, with strong ML Ops practices (MLflow, Docker, REST APIs, and AWS). My work consistently improves efficiency and speeds up processing for high-volume systems handling millions of records.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

English

Fluent

Work Experience

AI/ML Engineer at Kroger

April 1, 2025 - Present

Designed and deployed Generative AI solutions using LLMs to automate document understanding for merchandising and supplier workflows, reducing turnaround time from days to same-day processing. Architected a Semantic Product Intelligence System using Retrieval-Augmented Generation (RAG) and a vector store to unify search across product catalogs, enabling seconds-level answers and a 45% speed improvement in information retrieval. Developed LangChain-based orchestration workflows integrating LLMs with internal SQL systems and APIs, streamlining data access. Trained and deployed PyTorch-based recommendation/ranking models to enhance product discovery with reduced inference latency. Established prompt engineering practices and reusable templates, contributing to $70K in annual operational savings. Fine-tuned LLMs using LoRA/PEFT to build a Retail Compliance Assistant, delivering instant, context-aware responses and reducing manual review time by 30+ hours per week.

AI/ML Engineer at JPMorgan Chase

November 1, 2023 - March 31, 2025

Engineered a fraud detection system using XGBoost and logistic regression over 5M+ records, reducing false positives by 28%. Delivered a credit risk prediction model using ensemble methods, improving decision consistency and increasing approval accuracy by 1.3x. Architected an NLP-based document processing pipeline with Hugging Face transformers to extract key financial data from 10K+ documents per month, reducing manual review by 40%. Implemented customer segmentation using K-Means and PCA to identify high-value and at-risk users, enabling targeted campaigns that improved response rates by 18%. Deployed models as REST APIs using Flask and Docker on AWS EC2, handling 2K+ daily inferences with sub-120 ms latency. Standardized ML lifecycle management using MLflow, reducing deployment cycles from 2 weeks to 4 days.

Machine Learning Scientist at TCS

January 1, 2021 - July 1, 2022

Deployed DocIntel, an NLP pipeline using BERT and GPT models to extract and classify invoice and contract data, reducing document processing time from 3–4 hours to under 10 minutes (95% reduction) and improving downstream data accuracy for ETL workflows. Architected and implemented a RAG-based knowledge retrieval solution by integrating vector search with internal data sources, improving answer relevance and reducing manual search effort for business users by ~40%. Led end-to-end model development for classification and prediction use cases, achieving ~15% improvement over baseline. Partnered with data engineering to build scalable ETL pipelines orchestrated with Apache Airflow for structured and unstructured data (10M+ records), improving data readiness and reducing pre-processing time for analytics and ML workflows.