Summary
Overview
Work History
Education
Skills
Websites
Accomplishments
Certification
Timeline
Generic

Nandhakumar Kandasamy

Hyderabad

Summary

AI/ML professional with over 4 years of experience specializing in Large Language Models and Natural Language Processing. Proven expertise in developing Agentic Systems, Object Detection, and Reinforcement Learning. Committed to driving innovation and delivering impactful solutions in technology projects.

Overview

4
4
years of professional experience
1
1
Certification

Work History

Machine Learning Engineer

ServiceNow India
Hyderabad
06.2024 - Current
  • Involved in evaluating web agents by developing a unified benchmarking framework, creating supplementary datasets, and reproducing advanced workflows, while keeping up with the latest advancements in the field.
  • Enhanced NowLLM's robustness in agentic pipelines and general mathematical reasoning by generating high-quality datasets for supervised fine-tuning, alignment, and occasionally reinforcement learning with verifiable rewards; conducted data quality analysis, and trained models to evaluate data impact.
  • Conducted Red Teaming on LLMs by implementing direct and indirect attack strategies to probe ethical guideline violations; leveraged insights from academic research, and integrated open-source tools for adversarial testing.
  • Improved NowLLM's robustness against Red Teaming attacks by generating high-quality synthetic datasets for supervised fine-tuning and alignment, enhancing the model's resistance to adversarial prompts.

Machine Learning Engineer II

Factset Systems
Hyderabad
06.2021 - 06.2024
  • Developed and implemented models for Document Layout Analysis, resulting in a 30% reduction in manual intervention during table collection from filings.
  • Implemented a RAG-based system for content generation from summary notes using LlamaIndex.
  • Designed and built a Relevant Document Classifier to filter content from various sources, eliminating duplicates and increasing overall efficiency by 50%.
  • Successfully trained a Transformer-based Question Answering model for specific entity extraction, achieving over 80% accuracy for all entity types.
  • Engineered a Summarization model for a business use-case, automating hundreds of hours of manual effort with over 75% acceptable summaries across the target universe.
  • Implemented research papers to address various problems.
  • Integration/Continuous Deployment (CI/CD) pipelines, and perform low-cost inference using Lambda for model inferencing.
  • Experienced in developing Swagger APIs using Python frameworks.

Education

B.Tech - CSE

Indian Institute of Technology
Tirupati
05.2021

Skills

  • Natural language processing
  • Document layout analysis
  • Object detection
  • LangChain and LangGraph
  • Agentic-RAG methodology
  • Deep learning techniques
  • Python programming
  • SQL database management
  • PyTorch framework
  • OpenRLHF and Axolotl LLM training
  • Docker containerization
  • AWS cloud services
  • Hugging Face tools
  • Detectron2 framework

Websites

Accomplishments

  • Llama Award, 01/01/25, ServiceNow
  • Blue Ribbon Award, 01/01/23, Factset
  • Second Runner Up, 01/01/19, AI Hackathon, CDAC-Pune conducted by NVIDIA
  • First Prize, 01/01/19, Smart India Hackathon SW Edition

Certification

  • Generative AI with large language models - Deeplearning.AI
  • Building Multimodal Search and RAG - Deeplearning.AI

Timeline

Machine Learning Engineer

ServiceNow India
06.2024 - Current

Machine Learning Engineer II

Factset Systems
06.2021 - 06.2024

B.Tech - CSE

Indian Institute of Technology

Similar Profiles

CREATE PROFILE
Nandhakumar Kandasamy