Sakshi Verma

I’m an AI/ML engineer focused on designing and shipping real-world intelligent systems, with hands-on experience across NLP, RAG pipelines, and full-stack deployment. I’ve built production-grade solutions using BERT-family models, LangChain, FastAPI, and Docker, turning model outputs into usable applications.

In my AI/ML Engineer Intern role, I built “Resonova,” an emotion-preserving AI dubbing pipeline that converts English video to Hindi using Whisper ASR, IndicTrans2, XTTS-v2 voice cloning, and Wav2Lip. I also engineered a prosody-preservation layer (pitch, energy, speaking rate via librosa) to condition cloned speech, achieving >0.80 speaker similarity versus a ~0.50 baseline, then containerized and deployed a public demo on HuggingFace Spaces with an evaluation suite.

Alongside that, I’ve worked on data engineering and applied analytics to drive decisions—building structured marketing data pipelines, running funnel analysis, and creating a Power BI dashboard that exposed a 99% lead drop-off rate. In projects like PoetryDNA and HireIQ, I fine-tuned and combined models for classification and explainable retrieval, and I implemented RAG workflows to generate role-specific interview questions that target each candidate’s weak areas.