Looking for a job

Kajal Kumari

@kajalkumari10

Message

AI Engineer at Vecmocon Technologies who built a RAG document Q&A chatbot and helped cut EMI defaults by 20%.

India

Message

At Vecmocon Technologies, I architected a production RAG document Q&A chatbot using LangChain, LangGraph, ChromaDB, and GPT-4, then extended it to answer natural-language SQL queries with context awareness.

I also designed an Isolation Forest risk-profiling model that cut EMI default rates by 20%, and built real-time BaaS data ingestion pipelines on MongoDB and GCP.

At Zee Media, I designed and ran an autopay A/B test that lifted trial-to-paid conversion from 4% to 7%. I also migrated data pipelines from AWS to GCP and built dashboards for product, revenue, and marketing analysis.

Earlier, as a Junior Data Scientist at Curl Analytics, I built and productionized a document classification model that reached 95% accuracy and reduced processing latency from five minutes to three seconds. I’m an IIT Bombay graduate and a KVPY and NTSE Scholar.

Experience

Work history, roles, and key accomplishments

Lead Product Analyst

Zee Media

May 2026 - Sep 2026 (4 months)

Diagnosed payment-gateway autopay deduction limitations and ran an A/B test that lifted trial-to-paid conversion from 4% to 7%. Migrated data pipelines from AWS to GCP and built revenue, product-growth, and marketing-campaign dashboards.

A B Testing Product Analytics CleverTap Appsflyer Mixpanel Firebase Google Analytics NSAW KPI dashboarding AWS GCP

AI Engineer / Data Scientist

Vecmocon Technologies

Nov 2024 - Apr 2026 (1 year 5 months)

Architected a production RAG document Q&A chatbot with LangChain, LangGraph, and ChromaDB, and extended it to answer natural-language SQL queries. Built a Random Forest income-prediction model and an Isolation Forest risk-profiling model that cut EMI default rates by 20%.

RAG Lang Chain LangGraph ChromaDB GPT 4 SQL Random Forest Isolation Forest Docker MongoDB GCP

AI Engineer Intern

Trikl AI

Jun 2024 - Oct 2024 (4 months)

Built LLM pipelines for generating personalized LinkedIn content at scale and designed a RAG system with Pinecone and LLM for semantic search across user profiles.

GPT 4 Prompt Engineering RAG Pinecone Text Embeddings Semantic Search

Junior Data Scientist

Curl Analytics

Jul 2019 - Dec 2020 (1 year 5 months)

Built and productionized a document classification model achieving 95% accuracy and reduced processing latency from 5 minutes to 3 seconds via NumPy vectorization. Fine-tuned spaCy NER models and built a cosine-similarity embedding system for automatic document clustering.

Random Forest NumPy SpaCy NER Cosine Similarity scikit learn GitHub