Looking for a job

Vaidehi Pandey

@vaidehipandey

Senior Data Scientist at Qualys building multi-agent security query systems, including a natural-language-to-QQL engine.

India

Message

At Qualys, I led the architecture and development of a natural-language-to-QQL engine for generating schema-aware security queries. I also built MCP servers for Qualys products and integrated them into a LangGraph multi-agent framework.

I distilled LLaMA 70B orchestration logic into Phi-4-mini and deployed it with vLLM. I also built a semantic cache to reduce redundant LLM calls and latency for repeated or near-duplicate queries.

At Qualys, I fine-tuned T5-Flan-Small for QQL generation and developed a verification module that asks clarifying questions. I also built a multi-LLM chatbot that translates natural-language queries into Elasticsearch DSL and deployed its API on Kubernetes.

Previously, at Resleeve, I deployed a fine-tuned Stable Diffusion model for virtual try-on, with CI/CD to GCP for production inference. At Arspectra, I trained and integrated gesture-recognition models; at Capgemini, I trained classifiers for an NLP ticket-assignment tool.

Experience

Work history, roles, and key accomplishments

Current

Senior Data Scientist

Current

Qualys

Apr 2024 - Present (2 years 7 months)

Led architecture and development of a Natural Language to QQL translation engine and built MCP servers for multiple Qualys products. Integrated NLQ-QQL engine and product MCPs into a LangGraph-based multi-agent framework orchestrated by LLaMA 70B, and distilled LLaMA 70B into Phi-4-mini for faster inference.

Machine Learning Engineer

Resleeve

Sep 2022 - Jan 2024 (1 year 4 months)

Collected and validated large datasets for GAN-based virtual try-on, improving model accuracy and realism. Deployed a fine-tuned Stable Diffusion model with automated CI/CD pipeline and developed production inference scripts for text-to-image and image-to-image generation.

Gan Stable Diffusion Weights Biases GCP CI CD Docker Kubernetes Python PyTorch Data Collection Data Validation

Junior Machine Learning Engineer

Arspectra

Jan 2022 - Jun 2022 (5 months)

Curated datasets for hand detection, gesture recognition, and hand pose estimation, and trained and integrated models into an end-to-end pipeline. Conducted human evaluation of accuracy and responsiveness in real-time scenarios.

Hand Detection Gesture Recognition Human Hand Pose Python PyTorch Data Curation Model Training Integration Human Evaluation

Associate Consultant

Capgemini

Nov 2016 - Jan 2020 (3 years 2 months)

Collected and preprocessed ticket data to build a robust training set. Trained logistic regression, random forest, and SVM classifiers for automated ticket assignment, achieving an F1-score of 0.75.

Logistic Regression Random Forest SVM NLP data preprocessing Python scikit learn