arun kumar

AI / ML Engineer

2+ yrs exp 84 · Excellent

About

AI/ML Engineer with hands-on experience building and deploying production-grade Generative AI, Voice AI, and ML systems. Experienced in developing enterprise RAG pipelines, fine-tuning LLMs, and building conversational AI solutions used in real-world customer support environments. At Kapture CX, I have contributed to end-to-end AI systems including Voice AI agents, GPT-4o-based conversation quality analysis, AI-powered agent assist features, and speech recognition models. Passionate about solving complex engineering challenges and building scalable AI solutions that create measurable business impact.

Skills & Expertise (20)

LORA Expert

9.0/10

Years Exp

RAG Pipelines Expert

9.0/10

Years Exp

PyTorch Advanced

8.0/10

Years Exp

LangChain Advanced

8.0/10

Years Exp

Deep Learning Advanced

8.0/10

Years Exp

Prompt Engineering Advanced

8.0/10

Years Exp

Transformer Advanced

8.0/10

Years Exp

Hugging Face Advanced

8.0/10

Years Exp

Python Advanced

8.0/10

Years Exp

TensorFlow Intermediate

7.0/10

Years Exp

NumPy Intermediate

7.0/10

Years Exp

Pandas Intermediate

7.0/10

Years Exp

scikit-learn Intermediate

7.0/10

Years Exp

FastAPI Intermediate

7.0/10

Years Exp

GCP Intermediate

6.0/10

Years Exp

LangGraph Intermediate

6.0/10

Years Exp

Langsmith Intermediate

6.0/10

Years Exp

SQL Intermediate

6.0/10

Years Exp

Docker Intermediate

5.5/10

Years Exp

MCP Intermediate

5.0/10

Years Exp

Work Experience

ML Engineer

Kapture CX

Jun 2025 - Present

Engineered a real-time Voice AI agent for customer support by designing ASR pipelines, implementing VAD-based interruption handling and turn-taking mechanisms, and orchestrating LLM-driven conversational workflows to enable natural, human-like interactions. Fine-tuned Gemma 4 E4B IT for Tamil speech recognition using PeFT and LoRA techniques by reducing the Word Error Rate (WER) from 40% to 26% and successfully deploying the model into the production ASR pipeline. Built production-grade Retrieval Augmented Generation (RAG) systems using hybrid retrieval, vector databases, and re-ranking strategies to deliver accurate, context-aware knowledge retrieval for customer support workflows. Fine-tuned GPT-4o on labeled customer support conversations to automate conversation quality analysis, achieving approximately 91% agreement with human evaluators and significantly reducing manual QA effort.

ML Intern

Kapture CX

Jul 2024 - May 2025

Developed and deployed AI-powered agent assist capabilities including email and chat response recommendations, grammar correction, tone transformation, text expansion, and conversation summarization. Researched and implemented an audio denoising pipeline to suppress non-human background noise before transcription, improving ASR robustness and transcription accuracy for voice agents. Designed and maintained scalable FastAPI based backend services supporting Voice AI and Generative AI applications, ensuring reliable and efficient model inference in production environments.