Headline

Developer @ Genpact | Building AI Solutions for Automation

About

GenAI specialist at Genpact skilled in LLM fine-tuning, Multimodal AI, LangChain and RAG pipelines, and deploying models on AWS, Azure, and GCP- VertexAi.

Location

Bengaluru

AI stack

claude code

hugging face

langchain

langgraph

openai

Work experience

Agentic AI - Developer at GENPACT, Dec 2025 - Present · 10 mos, * Architected and shipped a multi-agent system suite (Semantic Kernel) with 4–5 use cases in production, serving 200+ analysts across 4+ enterprise workflows through autonomous task decomposition and tool routing * Designed a self-hosted LLM observability stack (Grafana, Loki, Prometheus, Tempo) on Azure infra, cutting debugging time 60% and incident response 40%

Data Scientist at GENPACT, Aug 2024 - Present · 2 yrs 2 mos, Bengaluru, * Data Scientist (Aug 2024 – Present) 📄 IEEE Published Research — Fine-tuned a Tiny Language Model using QLoRA (PEFT) to automate end-to-end insurance claims processing, reducing dependency on large-scale LLMs while maintaining high accuracy. Paper Link: https://ieeexplore.ieee.org/abstract/document/11472416 * Built a production async legal contract extraction pipeline (Textract → Claude 3.7 → Salesforce) with a caching layer to reduce both individual deal and batch processing time, processing 8,000+ contracts/week at 93.64% accuracy and cutting deal entry time 87.5% — scaled across 117 fields spanning 30+ regions/languages, each with distinct parsing rules and legal contract logic

Data Science Intern at GENPACT, Jan 2024 - Jul 2024 · 7 mos, Bengaluru, * Developed an open-source segmentation tool for medical applications, with a Streamlit front-end to assess and showcase model performance * Conducted a comprehensive comparison of state-of-the-art image-captioning models across relevant metrics, with findings applied to later projects * Built an end-to-end QnA application using RAGAS and LLMs to retrieve text and image data in structured form from user-provided contexts * Built a GPT-4 Vision + U-Net dual-model system for invoice fraud detection, achieving 90% accuracy with 30% fewer false positives, deployed on GCP Vertex AI

Data analytics intern at Edulyt India at Edulyt India, Aug 2023 - Sep 2023 · 2 mos, Remote

Machine Learning Intern at Suvidha Foundation (Suvidha Mahila Mandal), Aug 2023 - Sep 2023 · 2 mos

Data Science & Business Analytics at The Sparks Foundation, Jan 2022 - Jun 2022 · 6 mos

Education

Bachelor of Technology - BTech, Data Science, NIIT University

DY Patil University

Public links

youtube.com: https://youtube.com/@@omgholap86